feat: 根据激光找出图片中心点坐标

.idea/.gitignore

@@ -0,0 +1,8 @@
+# 默认忽略的文件
+/shelf/
+/workspace.xml
+# 基于编辑器的 HTTP 客户端请求
+/httpRequests/
+# Datasource local storage ignored files
+/dataSources/
+/dataSources.local.xml

.idea/.name

@@ -0,0 +1 @@
+network.py

.idea/archery.iml

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<module version="4">
+  <component name="PyDocumentationSettings">
+    <option name="format" value="PLAIN" />
+    <option name="myDocStringFormat" value="Plain" />
+  </component>
+</module>

.idea/inspectionProfiles/profiles_settings.xml

@@ -0,0 +1,6 @@
+<component name="InspectionProjectProfileManager">
+  <settings>
+    <option name="USE_PROJECT_PROFILE" value="false" />
+    <version value="1.0" />
+  </settings>
+</component>

.idea/misc.xml

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="Black">
+    <option name="sdkName" value="Python 3.13 virtualenv at H:\iot\racingiot_v1\.venv" />
+  </component>
+  <component name="ProjectRootManager" version="2" project-jdk-name="maixcam" project-jdk-type="Python SDK" />
+</project>

.idea/vcs.xml

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="VcsDirectoryMappings">
+    <mapping directory="" vcs="Git" />
+  </component>
+</project>

Untitled

@@ -0,0 +1 @@
+v1.2.15.1] [ERROR] main.py:416 - [MAIN] 显示异常: 'LaserManager' object has no attribute 'remote_detect_tick'

adc.py

@@ -4,12 +4,12 @@ from maix import time
 a = adc.ADC(0, adc.RES_BIT_12)
 
 while True:
-    # raw_data = a.read()
+    raw_data = a.read()
-    # print(f"ADC raw data:{raw_data}")
+    print(f"ADC raw data:{raw_data}")
-    # if raw_data > 2450:
+    if raw_data > 2450:
-    #     print(f"ADC raw data:{raw_data}")
+        print(f"ADC raw data:{raw_data}")
-    # elif raw_data < 2000:
+    elif raw_data < 2000:
-    #     print(f"ADC raw data:{raw_data}")
+        print(f"ADC raw data:{raw_data}")
     time.sleep_ms(1)
 
     vol = int(a.read_vol() * 10) / 10

app.yaml

@@ -1,6 +1,6 @@
 id: t11
 name: t11
-version: 2.15.15
+version: 2.1.1
 author: t11
 icon: ''
 desc: t11
@@ -14,17 +14,16 @@ files:
   - cameraParameters.xml
   - config.py
   - hardware.py
-  - laser_detector.py
   - laser_manager.py
   - logger_manager.py
   - main.py
   - model_270139.cvimodel
   - model_270139.mud
   - network.py
-  - ota_curl.sh
   - ota_manager.py
   - power.py
   - server.pem
+  - set_autostart.py
   - shoot_manager.py
   - shot_id_generator.py
   - target_roi_yolo.py

config.py

@@ -106,6 +106,14 @@ DEFAULT_LASER_POINT =  (320, 245)   # 默认激光中心点
 HARDCODE_LASER_POINT = True  # 是否使用硬编码的激光点（True=使用硬编码值，False=使用校准值）
 HARDCODE_LASER_POINT_VALUE = (320, 296) # 硬编码的激光点坐标(315, 245) #   # 硬编码的激光点坐标 (x, y)
 
+# 远程激光点识别（TCP cmd=200）：画面内找红点，稳定 N 秒且无明显跳动后上报坐标
+LASER_REMOTE_DETECT_STABLE_SEC = 3.0       # 连续稳定时长（秒）
+LASER_REMOTE_DETECT_MAX_MOVE_PX = 12.0     # 窗口内最大位移超过此值视为大幅移动，重新计时
+LASER_REMOTE_DETECT_SAMPLE_MS = 80         # 采样间隔
+LASER_REMOTE_DETECT_MIN_SAMPLES = 8        # 判定稳定前窗口内最少样本数
+LASER_REMOTE_DETECT_WARMUP_MS = 500        # cmd=200 开激光后等待稳定再采样
+# 远程识别会话无总超时：cmd=200 启动后持续检测并上报，直至 cmd=201 停止
+
 # 激光点检测配置
 LASER_DETECTION_THRESHOLD = 140  # 红色通道阈值（默认120，可调整，范围建议：100-150）
 LASER_RED_RATIO = 1.5  # 红色相对于绿色/蓝色的倍数要求（默认1.5，可调整，范围建议：1.3-2.0）
@@ -134,7 +142,7 @@ IMAGE_CENTER_Y = 240  # 图像中心 Y 坐标
 # ==================== 三角形四角标记：单应性偏移 + PnP 估距 ====================
 # 依赖 cameraParameters.xml（相机内参）与 triangle_positions.json（四角物方坐标，厘米或毫米见 JSON 约定）。
 # 部署时请把这两个文件放到 APP_DIR（与 main 同应用目录），或改下面路径为设备上的实际绝对路径。
-USE_TRIANGLE_OFFSET = False  # False 时仅走黄心圆/椭圆 + 半径估距，不使用三角形路径
+USE_TRIANGLE_OFFSET = True  # False 时仅走黄心圆/椭圆 + 半径估距，不使用三角形路径
 CAMERA_CALIB_XML = APP_DIR + "/cameraParameters.xml"
 TRIANGLE_POSITIONS_JSON = APP_DIR + "/triangle_positions.json"
 # 检测到的三角形边长在图像中的像素范围，分辨率或靶纸占比变化时可微调
@@ -144,6 +152,13 @@ TRIANGLE_SIZE_RANGE = (8, 500)
 # 如果射箭距离很固定，可设具体范围（如 min=2.5, max=6.0）作为额外保险
 TRIANGLE_DISTANCE_MIN_M = 0.0   # 0=不启用下限检查
 TRIANGLE_DISTANCE_MAX_M = 0.0   # 0=不启用上限检查
+# 三角形方向校验：四角黑三角应为 ◤ ◥ / ◣ ◢，即三角形从外角指向靶心；用于过滤相邻靶混入/跨靶组合
+TRIANGLE_DIRECTION_VALIDATE_ENABLE = False
+TRIANGLE_DIRECTION_MIN_PASS = 3  # 至少多少个真实三角方向正确才认为该组有效；3点补全时推荐3，误检多可设2
+TRIANGLE_DIRECTION_DOT_MIN = 0.0  # 方向点积阈值；0=只要求同向半平面，0.35≈夹角<70°，0.5≈夹角<60°
+TRIANGLE_DIRECTION_TO_CENTER_DOT_MIN = 0.35  # 必须指向候选靶心；0.35≈夹角<70°，用于过滤相邻靶混入
+TRIANGLE_CENTER_DISTANCE_VALIDATE_ENABLE = True  # 四角三角到候选靶心距离需近似一致，过滤跨靶组合
+TRIANGLE_CENTER_DISTANCE_RATIO_TOL = 0.45  # (max_dist-min_dist)/mean_dist 最大允许值；越小越严格
 # 三角形检测兜底增强：CLAHE（更鲁棒但更慢）。颜色阈值修复后通常不需要，保持关闭以优先速度。
 TRIANGLE_ENABLE_CLAHE_FALLBACK = False
 # 三角形检测调试：保存 Otsu 二值化图像（临时调试用，定位后关闭）
@@ -169,6 +184,7 @@ TRIANGLE_SHAPE_COS_TOLERANCE = 0.25    # 直角余弦绝对值上限（原 0.20
 # 建议设为实测最坏耗时的 1.2 倍；超时后圆心检测仍会并行跑完，跑完后若三角形已结束则优先用三角形。
 TRIANGLE_TIMEOUT_MS = 1000
 # True=打印各阶段耗时(ms)，用于定位瓶颈；稳定后可 False 减少日志
+ARCHERY_TIMING_ENABLE = False  # 总开关：False 关闭所有算法耗时统计（shoot_manager + triangle_target + vision）
 TRIANGLE_TIMING_LOG = True
 # True=Stage2 每个子框内传统三角失败时打一条统计（Otsu/Adaptive 下轮廓数与各拒绝原因计数）
 TRIANGLE_LOG_STAGE2_PATCH_REJECT = True
@@ -256,11 +272,15 @@ TRIANGLE_CROP_ROI_MIN_SIDE_PX = 64
 # 射箭保存图 / 预览上绘制 YOLO 靶环 ROI 矩形 (x0,y0,x1,y1)，核对是否裁准；不需要时改 False
 TRIANGLE_YOLO_DRAW_ROI_ON_SHOT = True
 # 物方采样调试：以靶心为中心，取半径 15cm 的圆周样本点，用于黑/白颜色对比
+TRIANGLE_SAMPLE_ENABLE = True
+TRIANGLE_SAMPLE_TIMING_ENABLE = True  # 仅统计物方采样耗时（其他 timing 可关）
 TRIANGLE_SAMPLE_RADIUS_CM = 15.0
 TRIANGLE_SAMPLE_ANGLES_DEG = (0, 90, 180, 270)
 TRIANGLE_SAMPLE_PATCH_HALF_PX = 2
+# 物方采样判断黑白阈值（R/G/B 均小于此值视为黑）；40cm 黑靶在靶面位置全黑，20cm 白靶则 R/G/B 偏高
+TRIANGLE_SAMPLE_BLACK_THRESH = 30.0
 # 开机阶段预加载 YOLO detector；detect 使用 dual_buff=False，避免返回上一帧结果。
-TRIANGLE_YOLO_PRELOAD_ON_BOOT = False
+TRIANGLE_YOLO_PRELOAD_ON_BOOT = True
 
 # ── 第二段 YOLO：仅在 Stage1 裁切出的靶环图上推理（与合成 stage2 训练数据一致）→ 子框内传统算法取直角点 ──
 # Stage1 靶环裁切内如何找黑三角标记（对比耗时时可切换）：
@@ -308,21 +328,16 @@ LASER_COLOR = (0, 255, 0)  # RGB颜色
 LASER_THICKNESS = 1
 LASER_LENGTH = 2
 
-# ==================== 队列大小限制（防止内存泄漏） ====================
-MAX_SEND_QUEUE_SIZE = 500       # 发送队列上限
-MAX_TCP_PAYLOADS = 500          # AT TCP 载荷缓存上限
-MAX_HTTP_EVENTS = 200           # AT HTTP 事件缓存上限
-LOG_QUEUE_MAXSIZE = 10000       # 日志队列上限
-MAX_CMD_THREADS = 10            # 并发命令线程上限（防止服务器下发命令时无限创建线程）
-
 # ==================== 图像保存配置 ====================
-SAVE_IMAGE_ENABLED = False  # 是否保存图像（True=保存，False=不保存）
+SAVE_IMAGE_ENABLED = True  # 是否保存图像（True=保存，False=不保存）
+SAVE_RAW_SHOT_IMAGE_ENABLED = False  # 是否额外保存射箭原图；可通过 TCP cmd=46 动态开关
+VISION_TIMING_ENABLE = True  # 视觉圆检测耗时统计（detect_circle_v3 内部各步骤耗时）
 PHOTO_DIR = "/root/phot"    # 照片存储目录
 MAX_IMAGES = 1000
 # Stage2 调试目录（默认 PHOTO_DIR/stage2_roi）内 JPEG 最多保留张数；None 表示与 MAX_IMAGES 相同
 TRIANGLE_BLACK_YOLO_STAGE2_ROI_MAX_IMAGES = None
 
-SHOW_CAMERA_PHOTO_WHILE_SHOOTING = False  # 是否在拍摄时显示摄像头图像（True=显示，False=不显示），建议在连着USB测试过程中打开
+SHOW_CAMERA_PHOTO_WHILE_SHOOTING = True  # 是否在拍摄时显示摄像头图像（True=显示，False=不显示），建议在连着USB测试过程中打开
 
 # ==================== OTA配置 ====================
 MAX_BACKUPS = 5

laser_detector.py

@@ -1,248 +0,0 @@
-from maix import image, time
-from logger_manager import logger_manager
-from camera_manager import camera_manager
-
-_USE_CV = False
-try:
-    import cv2
-    import numpy as np
-
-    _USE_CV = True
-except ImportError:
-    pass
-
-WIDTH = 640
-HEIGHT = 480
-THRESHOLD = 100
-RED_RATIO = 1.5
-SEARCH_RADIUS = 80
-TRACK_RADIUS = 30
-MIN_PIXELS = 3
-COARSE_STEP = 2
-STABLE_COUNT = 2
-MAX_SKIP_FRAMES = 5
-
-# Temporal smoothing
-_EMA_ALPHA = 0.35
-_GATE_PX = 10
-_FRAME_INTERVAL_MS = 50
-
-_prev_smoothed = None
-
-
-def _red_weighted_centroid(r_ch, g_ch, b_ch, mask, x0, y0):
-    y_ids, x_ids = np.where(mask)
-    if len(y_ids) == 0:
-        return None
-    r_vals = r_ch[y_ids, x_ids].astype(np.float64)
-    g_vals = g_ch[y_ids, x_ids].astype(np.float64)
-    b_vals = b_ch[y_ids, x_ids].astype(np.float64)
-    w = r_vals - np.maximum(g_vals, b_vals)
-    w = np.clip(w, 0, None)
-    w = w * w
-    total_w = w.sum()
-    if total_w < 1e-6:
-        return None
-    cx = (x_ids.astype(np.float64) * w).sum() / total_w + x0
-    cy = (y_ids.astype(np.float64) * w).sum() / total_w + y0
-    return (float(cx), float(cy))
-
-
-def find_ellipse(img_cv, cx, cy, roi_r, th, ratio):
-    x1 = max(0, cx - roi_r)
-    x2 = min(WIDTH, cx + roi_r)
-    y1 = max(0, cy - roi_r)
-    y2 = min(HEIGHT, cy + roi_r)
-    roi = img_cv[y1:y2, x1:x2]
-    if roi.size == 0:
-        return None
-    r = roi[:, :, 0].astype(np.int32)
-    g = roi[:, :, 1].astype(np.int32)
-    b = roi[:, :, 2].astype(np.int32)
-    mask = (r > th) & (r > g * ratio) & (r > b * ratio)
-    oe = (r > 200) & (g > 200) & (b > 200) & (r >= g) & (r >= b) & ((r - g) > 10) & ((r - b) > 10)
-    combined = (mask | oe).astype(np.uint8) * 255
-    contours, _ = cv2.findContours(combined, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-    if not contours:
-        return None
-    largest = max(contours, key=cv2.contourArea)
-    if cv2.contourArea(largest) < 5:
-        return None
-    cnt = largest.copy()
-    for pt in cnt:
-        pt[0][0] += x1
-        pt[0][1] += y1
-    ellipse_valid = len(cnt) >= 5
-    if ellipse_valid:
-        (ex, ey), (ew, eh), ang = cv2.fitEllipse(cnt)
-        mask_ellipse = np.zeros((HEIGHT, WIDTH), dtype=np.uint8)
-        cv2.ellipse(mask_ellipse, (int(ex), int(ey)), (int(ew / 2), int(eh / 2)), ang, 0, 360, 255, -1)
-        return _red_weighted_centroid(
-            img_cv[:, :, 0], img_cv[:, :, 1], img_cv[:, :, 2],
-            mask_ellipse > 0, 0, 0
-        )
-    M = cv2.moments(cnt)
-    if M["m00"] > 0:
-        return (float(M["m10"] / M["m00"]), float(M["m01"] / M["m00"]))
-    return None
-
-
-def is_red(r, g, b, th, ratio):
-    if r > th and r > g * ratio and r > b * ratio:
-        return True
-    if (r > 200 and g > 200 and b > 200 and r >= g and r >= b
-            and (r - g) > 10 and (r - b) > 10):
-        return True
-    return False
-
-
-def find_brightest_bytes(frame, cx, cy, roi_r, th, ratio):
-    x1 = max(0, cx - roi_r)
-    x2 = min(WIDTH, cx + roi_r)
-    y1 = max(0, cy - roi_r)
-    y2 = min(HEIGHT, cy + roi_r)
-    data = frame.to_bytes()
-
-    best_score = 0
-    best_x = (x1 + x2) // 2
-    best_y = (y1 + y2) // 2
-    found_any = False
-    for y in range(y1, y2, COARSE_STEP):
-        for x in range(x1, x2, COARSE_STEP):
-            idx = (y * WIDTH + x) * 3
-            r = data[idx]
-            g = data[idx + 1]
-            b = data[idx + 2]
-            if is_red(r, g, b, th, ratio):
-                score = r + g + b
-                dx = x - cx
-                dy = y - cy
-                dist_decay = max(0.5, 1.0 - ((dx * dx + dy * dy) ** 0.5 / roi_r) * 0.5)
-                score *= dist_decay
-                if score > best_score:
-                    best_score = score
-                    best_x = x
-                    best_y = y
-                    found_any = True
-
-    if not found_any:
-        return None
-
-    sf = 4
-    fx1 = max(x1, best_x - sf)
-    fx2 = min(x2, best_x + sf + 1)
-    fy1 = max(y1, best_y - sf)
-    fy2 = min(y2, best_y + sf + 1)
-
-    sum_x = 0.0
-    sum_y = 0.0
-    total_w = 0.0
-    count = 0
-    for y in range(fy1, fy2):
-        for x in range(fx1, fx2):
-            idx = (y * WIDTH + x) * 3
-            r = data[idx]
-            g = data[idx + 1]
-            b = data[idx + 2]
-            if is_red(r, g, b, th, ratio):
-                w = r + g + b
-                sum_x += x * w
-                sum_y += y * w
-                total_w += w
-                count += 1
-
-    if count < MIN_PIXELS:
-        return (float(best_x), float(best_y))
-
-    return (float(sum_x / total_w), float(sum_y / total_w))
-
-
-def _ema_filter(pos, alpha=_EMA_ALPHA):
-    global _prev_smoothed
-    if _prev_smoothed is None:
-        _prev_smoothed = pos
-        return pos
-    sx = alpha * pos[0] + (1 - alpha) * _prev_smoothed[0]
-    sy = alpha * pos[1] + (1 - alpha) * _prev_smoothed[1]
-    _prev_smoothed = (sx, sy)
-    return _prev_smoothed
-
-
-def _gated(pos, gate_px=_GATE_PX):
-    global _prev_smoothed
-    if _prev_smoothed is None:
-        return True
-    dx = pos[0] - _prev_smoothed[0]
-    dy = pos[1] - _prev_smoothed[1]
-    return (dx * dx + dy * dy) <= gate_px * gate_px
-
-
-def get_stable_laser_point(timeout_ms=15000, stable_count=STABLE_COUNT):
-    global _prev_smoothed
-    _prev_smoothed = None
-    try:
-        last_raw = None
-        stable = 0
-        start = time.ticks_ms()
-        cx, cy = WIDTH // 2, HEIGHT // 2
-        track_count = 0
-        skip_count = 0
-        while True:
-            if abs(time.ticks_diff(time.ticks_ms(), start)) > timeout_ms:
-                _prev_smoothed = None
-                return None
-            frame = camera_manager.read_frame()
-            if frame is None:
-                time.sleep_ms(10)
-                continue
-
-            if track_count > 0 and _prev_smoothed is not None:
-                search_cx = int(_prev_smoothed[0])
-                search_cy = int(_prev_smoothed[1])
-                search_r = TRACK_RADIUS
-            else:
-                search_cx = cx
-                search_cy = cy
-                search_r = SEARCH_RADIUS
-
-            pos_bright = find_brightest_bytes(frame, search_cx, search_cy, search_r, THRESHOLD, RED_RATIO)
-            pos = pos_bright
-            if _USE_CV:
-                img_cv = image.image2cv(frame, False, False)
-                pos_ellipse = find_ellipse(img_cv, search_cx, search_cy, search_r, THRESHOLD, RED_RATIO)
-                if pos_ellipse is not None:
-                    pos = pos_ellipse
-
-            if pos is not None:
-                skip_count = 0
-                track_count += 1
-                filtered = _ema_filter(pos)
-                if last_raw is not None:
-                    dx = abs(filtered[0] - last_raw[0])
-                    dy = abs(filtered[1] - last_raw[1])
-                    if dx <= 2 and dy <= 2:
-                        stable += 1
-                    else:
-                        stable = 1
-                else:
-                    stable = 1
-                last_raw = filtered
-                if logger_manager.logger:
-                    logger_manager.logger.info(f"pos:{pos},filtered:{filtered},stable:{stable}")
-                if stable >= stable_count:
-                    result = (int(filtered[0]), int(filtered[1]))
-                    _prev_smoothed = None
-                    return result
-            else:
-                skip_count += 1
-                if logger_manager.logger:
-                    logger_manager.logger.info(f"find_brightest_bytes None, skip={skip_count}, track={track_count}, search_center=({search_cx},{search_cy}), search_r={search_r}")
-                if skip_count > MAX_SKIP_FRAMES:
-                    _prev_smoothed = None
-                    track_count = 0
-                    stable = 0
-                    last_raw = None
-
-            time.sleep_ms(_FRAME_INTERVAL_MS)
-    finally:
-        _prev_smoothed = None

laser_manager.py

@@ -6,6 +6,7 @@
 """
 import _thread
 import json
+import math
 import os
 import binascii
 from maix import time
@@ -34,9 +35,13 @@ class LaserManager:
         self._calibration_active = False
         self._calibration_result = None
         self._calibration_lock = threading.Lock()
+        self._remote_detect_active = False
+        self._remote_detect_lock = threading.Lock()
+        self._remote_detect_result = None
         self._laser_point = None
         self._laser_turned_on = False
         self._last_frame_with_ellipse = None  # 保存绘制了椭圆的图像（用于调试/显示）
+        self._remote_detect_last_pos = None
         self._initialized = True
     
     # ==================== 状态访问（只读属性）====================
@@ -54,8 +59,8 @@ class LaserManager:
     @property
     def laser_point(self):
         """当前激光点（如果启用硬编码，则返回硬编码值）"""
-        # if config.HARDCODE_LASER_POINT:
+        if config.HARDCODE_LASER_POINT:
-        #     return config.HARDCODE_LASER_POINT_VALUE
+            return config.HARDCODE_LASER_POINT_VALUE
         return self._laser_point
     
     def get_last_frame_with_ellipse(self):
@@ -102,30 +107,262 @@ class LaserManager:
     # ==================== 业务方法 ====================
     
     def load_laser_point(self):
-        """加载激光中心点：优先使用本地保存的坐标，其次硬编码值，最后默认值"""
+        """从配置文件加载激光中心点，失败则使用默认值
-        # 优先：从本地持久化文件加载（由 cmd 201 保存）
+        如果启用硬编码模式，则直接使用硬编码值
+        """
+        if config.HARDCODE_LASER_POINT:
+            # 硬编码模式：直接使用硬编码值
+            self._laser_point = config.HARDCODE_LASER_POINT_VALUE
+            self.logger.info(f"[LASER] 使用硬编码激光点: {self._laser_point}")
+            return self._laser_point
+        
+        # 正常模式：从配置文件加载
         try:
-            if "laser_config.json" in os.listdir("/root"):
+            if os.path.exists(config.CONFIG_FILE):
                 with open(config.CONFIG_FILE, "r") as f:
                     data = json.load(f)
                     if isinstance(data, list) and len(data) == 2:
                         self._laser_point = (int(data[0]), int(data[1]))
-                        self.logger.info(f"[LASER] 从本地加载激光点: {self._laser_point}")
+                        self.logger.debug(f"[INFO] 加载激光点: {self._laser_point}")
                         return self._laser_point
-        except Exception:
+                    else:
-            pass
+                        raise ValueError
-
+            else:
-        # 其次：硬编码值
+                self._laser_point = config.DEFAULT_LASER_POINT
-        if config.HARDCODE_LASER_POINT:
+        except Exception as e:
-            self._laser_point = config.HARDCODE_LASER_POINT_VALUE
+            if self.logger:
-            self.logger.info(f"[LASER] 使用硬编码激光点: {self._laser_point}")
+                self.logger.warning(f"[LASER] 加载激光点失败，使用默认值: {e}")
-            return self._laser_point
+            self._laser_point = config.DEFAULT_LASER_POINT
-
+        
-        # 最后：默认值
-        self._laser_point = config.DEFAULT_LASER_POINT
-        self.logger.info(f"[LASER] 使用默认激光点: {self._laser_point}")
         return self._laser_point
     
+    @property
+    def remote_detect_active(self):
+        with self._remote_detect_lock:
+            return self._remote_detect_active
+
+    def get_remote_detect_result(self):
+        """获取并清除远程激光识别结果 (x, y) 或 None。"""
+        with self._remote_detect_lock:
+            result = self._remote_detect_result
+            self._remote_detect_result = None
+            return result
+
+    def remote_detect_tick(self, frame):
+        """
+        主循环显示路径调用的轻量 tick。
+        兼容旧调用点：当前远程识别由后台线程处理，这里不做重计算，
+        仅保留接口避免 AttributeError。
+        """
+        return None
+
+    def overlay_remote_detect_preview(self, frame):
+        """
+        在预览画面叠加远程识别点与坐标文本。
+        """
+        try:
+            import cv2
+            from maix import image
+            with self._remote_detect_lock:
+                pos = self._remote_detect_last_pos
+            if not pos:
+                return frame
+
+            img_cv = image.image2cv(frame, False, False)
+            if img_cv is None or img_cv.size == 0:
+                return frame
+
+            x, y = int(pos[0]), int(pos[1])
+            h, w = img_cv.shape[:2]
+            if x < 0 or y < 0 or x >= w or y >= h:
+                return frame
+
+            color = (255, 0, 0)  # RGB
+            cv2.circle(img_cv, (x, y), 8, color, 2)
+            cv2.line(img_cv, (x - 12, y), (x + 12, y), color, 1)
+            cv2.line(img_cv, (x, y - 12), (x, y + 12), color, 1)
+            cv2.putText(img_cv, f"laser=({x},{y})", (max(5, x + 10), max(20, y - 10)),
+                        cv2.FONT_HERSHEY_SIMPLEX, 0.55, color, 1, cv2.LINE_AA)
+
+            return image.cv2image(img_cv, False, False)
+        except Exception as e:
+            if self.logger:
+                self.logger.debug(f"[LASER-REMOTE] overlay 绘制失败: {e}")
+            return frame
+
+    def _set_remote_detect_result(self, result):
+        with self._remote_detect_lock:
+            self._remote_detect_result = result
+
+    def set_hardcoded_laser_point(self, x, y):
+        """更新 config.HARDCODE_LASER_POINT_VALUE（TCP cmd=201）。"""
+        try:
+            ix = int(round(float(x)))
+            iy = int(round(float(y)))
+        except (TypeError, ValueError) as e:
+            raise ValueError(f"invalid laser point ({x!r}, {y!r})") from e
+        config.HARDCODE_LASER_POINT = True
+        config.HARDCODE_LASER_POINT_VALUE = (ix, iy)
+        self._laser_point = (ix, iy)
+        try:
+            with open(config.CONFIG_FILE, "w") as f:
+                json.dump([ix, iy], f)
+        except Exception as e:
+            if self.logger:
+                self.logger.warning(f"[LASER] 保存硬编码激光点到本地失败: {e}")
+            raise
+        if self.logger:
+            self.logger.info(
+                f"[LASER] 已设置硬编码激光点 HARDCODE_LASER_POINT_VALUE=({ix}, {iy}) 并已保存到 {config.CONFIG_FILE}"
+            )
+        return ix, iy
+
+    def start_remote_laser_detect(self):
+        """
+        启动远程激光识别会话（TCP cmd=200）：开激光后持续检测。
+        每次稳定 3s 上报一次坐标，外循环直到 cmd=201 调用 stop_remote_laser_detect()。
+        Returns:
+            True 已启动；False 会话已在运行
+        """
+        with self._remote_detect_lock:
+            if self._remote_detect_active:
+                return False
+            self._remote_detect_active = True
+            self._remote_detect_result = None
+            self._remote_detect_last_pos = None
+        _thread.start_new_thread(self._remote_laser_detect_worker, ())
+        if self.logger:
+            self.logger.info("[LASER] 远程激光识别已启动 (cmd=200)")
+        return True
+
+    def stop_remote_laser_detect(self):
+        with self._remote_detect_lock:
+            self._remote_detect_active = False
+
+    def _remote_laser_detect_worker(self):
+        from camera_manager import camera_manager
+
+        stable_sec = float(getattr(config, "LASER_REMOTE_DETECT_STABLE_SEC", 3.0))
+        max_move = float(getattr(config, "LASER_REMOTE_DETECT_MAX_MOVE_PX", 12.0))
+        sample_ms = int(getattr(config, "LASER_REMOTE_DETECT_SAMPLE_MS", 80))
+        min_samples = int(getattr(config, "LASER_REMOTE_DETECT_MIN_SAMPLES", 8))
+        warmup_ms = int(getattr(config, "LASER_REMOTE_DETECT_WARMUP_MS", 500))
+        stable_ms = int(max(500, stable_sec * 1000))
+
+        samples = []
+        miss_count = 0
+        stable_hit_count = 0
+        reported = False
+
+        try:
+            if not self._laser_turned_on:
+                try:
+                    self.turn_on_laser()
+                except Exception as e:
+                    if self.logger:
+                        self.logger.warning(f"[LASER] cmd200 worker 开激光失败: {e}")
+            if warmup_ms > 0:
+                if self.logger:
+                    self.logger.info(f"[LASER] cmd200 激光预热 {warmup_ms}ms …")
+                time.sleep_ms(warmup_ms)
+
+            if self.logger:
+                self.logger.info("[LASER] 远程识别外循环已启动，直至 cmd=201 停止")
+
+            while True:
+                with self._remote_detect_lock:
+                    if not self._remote_detect_active:
+                        if self.logger:
+                            self.logger.info("[LASER] 远程识别会话结束 (cmd=201 或取消)")
+                        return
+
+                try:
+                    frame = camera_manager.read_frame()
+                    pos = self.find_red_laser_remote(frame)
+                except Exception as e:
+                    if self.logger:
+                        self.logger.warning(f"[LASER] 远程识别帧异常: {e}")
+                    pos = None
+                    time.sleep_ms(sample_ms)
+                    continue
+
+                now_ms = time.ticks_ms()
+                if pos is None:
+                    miss_count += 1
+                    samples.clear()
+                    stable_hit_count = 0
+                    if miss_count == 1 or miss_count % 40 == 0:
+                        if self.logger:
+                            self.logger.info(
+                                f"[LASER-REMOTE] 本帧未检出激光点（累计 {miss_count} 帧），"
+                                f"全图多策略搜索中…"
+                            )
+                    time.sleep_ms(sample_ms)
+                    continue
+
+                miss_count = 0
+                x, y = float(pos[0]), float(pos[1])
+                samples.append((now_ms, x, y))
+                cutoff = now_ms - stable_ms
+                samples = [(t, px, py) for t, px, py in samples if t >= cutoff]
+
+                if len(samples) < 2:
+                    time.sleep_ms(sample_ms)
+                    continue
+
+                xs = [s[1] for s in samples]
+                ys = [s[2] for s in samples]
+                span = max(
+                    max(xs) - min(xs),
+                    max(ys) - min(ys),
+                )
+                for i in range(len(samples)):
+                    for j in range(i + 1, len(samples)):
+                        d = math.hypot(
+                            samples[i][1] - samples[j][1],
+                            samples[i][2] - samples[j][2],
+                        )
+                        span = max(span, d)
+
+                if span > max_move:
+                    if self.logger:
+                        self.logger.debug(
+                            f"[LASER] 检测到大幅位移 span={span:.1f}px>{max_move}，重新计时"
+                        )
+                    samples.clear()
+                    stable_hit_count = 0
+                    time.sleep_ms(sample_ms)
+                    continue
+
+                window_ms = samples[-1][0] - samples[0][0]
+                if window_ms >= stable_ms and len(samples) >= min_samples:
+                    fx = int(round(sum(xs) / len(xs)))
+                    fy = int(round(sum(ys) / len(ys)))
+                    stable_hit_count += 1
+                    if self.logger:
+                        self.logger.info(
+                            f"[LASER] 远程识别稳定命中 {stable_hit_count}/3 span={span:.1f}px → ({fx}, {fy})"
+                        )
+                    samples.clear()
+                    if stable_hit_count >= 3 and not reported:
+                        reported = True
+                        self._set_remote_detect_result(
+                            {"result":"laser_detect_ok", "x": fx, "y": fy}
+                        )
+                        if self.logger:
+                            self.logger.info(
+                                f"[LASER] 已连续3次坐标稳定，完成上报，继续等待 cmd=201 关闭会话"
+                            )
+                    time.sleep_ms(sample_ms)
+                    continue
+
+                time.sleep_ms(sample_ms)
+        finally:
+            if self.logger:
+                self.logger.info("[LASER] 远程识别线程退出，等待下一次 cmd=200")
+            with self._remote_detect_lock:
+                self._remote_detect_active = False
+
     def save_laser_point(self, point):
         """保存激光中心点到配置文件
         如果启用硬编码模式，则不保存（直接返回 True）
@@ -828,6 +1065,172 @@ class LaserManager:
             # 使用原来的最亮点方法
             return self._find_red_laser_brightest(frame, threshold, search_radius, ellipse_params)
 
+    def find_red_laser_remote(self, frame):
+        """
+        cmd=200 远程识别专用：全图搜索、多策略、放宽阈值，不限距画面中心距离。
+        常规 find_red_laser 仅搜中心 ±LASER_SEARCH_RADIUS 且距中心 >50px 会丢弃。
+        """
+        import cv2
+        import numpy as np
+        from maix import image
+
+        img_cv = image.image2cv(frame, False, False)
+        if img_cv is None or img_cv.size == 0:
+            return None
+        h, w = img_cv.shape[:2]
+
+        r = img_cv[:, :, 0].astype(np.int32)
+        g = img_cv[:, :, 1].astype(np.int32)
+        b = img_cv[:, :, 2].astype(np.int32)
+        brightness = r + g + b
+        red_ratio = float(getattr(config, "LASER_RED_RATIO", 1.5))
+        ratio_lo = max(1.15, red_ratio - 0.35)
+
+        strategies = []
+        base_th = int(getattr(config, "LASER_DETECTION_THRESHOLD", 140))
+        for th in (base_th, 120, 100, 80, 60):
+            mask = (
+                (r > th)
+                & (r > g * ratio_lo)
+                & (r > b * ratio_lo)
+            )
+            strategies.append(("rgb", th, mask))
+
+        oe_th = int(getattr(config, "LASER_OVEREXPOSED_THRESHOLD", 200))
+        oe_diff = int(getattr(config, "LASER_OVEREXPOSED_DIFF", 10))
+        mask_oe = (
+            (r > oe_th - 30)
+            & (g > oe_th - 40)
+            & (b > oe_th - 40)
+            & (r >= g)
+            & (r >= b)
+            & ((r - g) > max(5, oe_diff - 5))
+            & ((r - b) > max(5, oe_diff - 5))
+        )
+        strategies.append(("overexposed", oe_th, mask_oe))
+
+        mask_bright = (brightness > 380) & (r >= g) & (r >= b) & ((r - g) > 3)
+        strategies.append(("bright", 0, mask_bright))
+
+        hsv = cv2.cvtColor(img_cv, cv2.COLOR_RGB2HSV)
+        hc, sc, vc = cv2.split(hsv)
+        mask_hsv = ((hc <= 18) | (hc >= 162)) & (sc >= 60) & (vc >= 60)
+        strategies.append(("hsv", 0, mask_hsv))
+
+        best_pos = None
+        best_score = -1.0
+        best_tag = None
+
+        max_area = float(getattr(config, "LASER_REMOTE_MAX_AREA", 300.0))
+        min_circularity = float(getattr(config, "LASER_REMOTE_MIN_CIRCULARITY", 0.25))
+
+        for name, th, mask in strategies:
+            m = (mask.astype(np.uint8)) * 255
+            if cv2.countNonZero(m) == 0:
+                continue
+            contours, _ = cv2.findContours(m, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+            if not contours:
+                continue
+            for cnt in contours:
+                area = cv2.contourArea(cnt)
+                if area < 1.5 or area > max_area:
+                    continue
+                peri = cv2.arcLength(cnt, True)
+                if peri <= 0:
+                    continue
+                circularity = float(4.0 * math.pi * area / (peri * peri))
+                if circularity < min_circularity:
+                    continue
+                M = cv2.moments(cnt)
+                if M["m00"] <= 0:
+                    continue
+                cx = float(M["m10"] / M["m00"])
+                cy = float(M["m01"] / M["m00"])
+                ix, iy = int(round(cx)), int(round(cy))
+                if ix < 0 or iy < 0 or ix >= w or iy >= h:
+                    continue
+                local_r = float(r[iy, ix])
+                score = area * local_r * (1.0 + local_r / 255.0) * (0.5 + circularity)
+                if score > best_score:
+                    best_score = score
+                    best_pos = (ix, iy)
+                    best_tag = (name, th, area)
+
+        if best_pos is not None:
+            with self._remote_detect_lock:
+                self._remote_detect_last_pos = best_pos
+            self._save_remote_detect_debug_image(frame, best_pos, best_tag)
+            if self.logger:
+                self.logger.info(
+                    f"[LASER-REMOTE] 检测到激光点 {best_pos} "
+                    f"strategy={best_tag[0]} th={best_tag[1]} area={best_tag[2]:.1f}"
+                )
+        elif self.logger:
+            self.logger.debug("[LASER-REMOTE] 未通过面积/圆度过滤")
+        return best_pos
+
+    def _save_remote_detect_debug_image(self, frame, pos, tag=None):
+        """保存远程识别调试图：叠加激光坐标并落盘。"""
+        try:
+            if not bool(getattr(config, "SAVE_IMAGE_ENABLED", True)):
+                return
+            import cv2
+            from maix import image
+
+            img_cv = image.image2cv(frame, False, False)
+            if img_cv is None or img_cv.size == 0:
+                return
+
+            x, y = int(pos[0]), int(pos[1])
+            h, w = img_cv.shape[:2]
+            if x < 0 or y < 0 or x >= w or y >= h:
+                return
+
+            cv2.circle(img_cv, (x, y), 8, (255, 0, 0), 2)
+            cv2.line(img_cv, (x - 12, y), (x + 12, y), (255, 0, 0), 1)
+            cv2.line(img_cv, (x, y - 12), (x, y + 12), (255, 0, 0), 1)
+
+            desc = ""
+            if tag:
+                desc = f" {tag[0]} th={tag[1]} area={tag[2]:.1f}"
+            cv2.putText(
+                img_cv,
+                f"laser=({x},{y}){desc}",
+                (10, 24),
+                cv2.FONT_HERSHEY_SIMPLEX,
+                0.55,
+                (255, 0, 0),
+                1,
+                cv2.LINE_AA,
+            )
+
+            base_dir = getattr(config, "PHOTO_DIR", "/root/phot")
+            debug_dir = f"{base_dir}/laser_remote"
+            try:
+                if debug_dir not in os.listdir("/root") and "/" not in debug_dir.replace("/root/", ""):
+                    os.mkdir(debug_dir)
+                else:
+                    try:
+                        os.makedirs(debug_dir, exist_ok=True)
+                    except Exception:
+                        pass
+            except Exception:
+                try:
+                    os.makedirs(debug_dir, exist_ok=True)
+                except Exception:
+                    return
+
+            ts = int(time.ticks_ms())
+            filename = f"{debug_dir}/remote_{x}_{y}_{ts}.jpg"
+            out = image.cv2image(img_cv, False, False)
+            out.save(filename)
+
+            if self.logger:
+                self.logger.info(f"[LASER-REMOTE] 调试图已保存: {filename}")
+        except Exception as e:
+            if self.logger:
+                self.logger.warning(f"[LASER-REMOTE] 保存调试图失败: {e}")
+
     def calibrate_laser_position(self, timeout_ms=8000, check_sharpness=True):
         """
         执行激光校准：循环拍照 → 检测靶心 → 检查激光点清晰度 → 找红点 → 保存坐标
@@ -1261,28 +1664,6 @@ class LaserManager:
                 except Exception as e:
                     self.logger.error(f"[LASER] 关闭激光失败: {e}")
 
-    def set_hardcoded_laser_point(self, raw_x, raw_y):
-        """
-        设置服务下发的硬编码激光点坐标，并保存到本地持久化文件。
-        下次启动时 load_laser_point() 会优先使用此保存的值。
-
-        Args:
-            raw_x: 服务下发的 x 坐标
-            raw_y: 服务下发的 y 坐标
-
-        Returns:
-            (int_x, int_y) 元组
-        """
-        ix = int(raw_x)
-        iy = int(raw_y)
-        self._laser_point = (ix, iy)
-        try:
-            with open(config.CONFIG_FILE, "w") as f:
-                json.dump([ix, iy], f)
-            self.logger.info(f"[LASER] 设置并持久化激光点: ({ix}, {iy})")
-        except Exception as e:
-            self.logger.error(f"[LASER] 持久化激光点失败: {e}")
-        return ix, iy
 
 # 创建全局单例实例
 laser_manager = LaserManager()

logger_manager.py

@@ -65,8 +65,8 @@ class LoggerManager:
             backup_count = config.LOG_BACKUP_COUNT
         
         try:
-            # 创建日志队列（有界队列，防止内存泄漏；满时自动丢弃旧日志）
+            # 创建日志队列（无界队列）
-            self._log_queue = queue.Queue(maxsize=config.LOG_QUEUE_MAXSIZE)
+            self._log_queue = queue.Queue(-1)
             
             # 确保日志文件所在的目录存在
             log_dir = os.path.dirname(log_file)

main.py

@@ -290,33 +290,34 @@ def cmd_str():
     last_avg_abs = 0
 
     def _flush_pressure_buf(reason: str):
+        if not config.AIR_PRESSURE_lOG:
+            return
         nonlocal pressure_buf, pressure_sum, pressure_min, pressure_max, pressure_t0_ms, logger, pressure_abs_sum, last_avg_abs
         if not pressure_buf:
             return
-        if config.AIR_PRESSURE_lOG:
+        t1_ms = time.ticks_ms()
-            t1_ms = time.ticks_ms()
+        n = len(pressure_buf)
-            n = len(pressure_buf)
+        avg = (pressure_sum / n) if n else 0
-            avg = (pressure_sum / n) if n else 0
+        avg_abs = (pressure_abs_sum / n) if n else 0
-            avg_abs = (pressure_abs_sum / n) if n else 0
+        # 一行输出：方便后处理画曲线；同时带上统计信息便于快速看波峰
-            line = (
+        line = (
-                f"[气压批量] reason={reason} "
+            f"[气压批量] reason={reason} "
-                f"t0={pressure_t0_ms} t1={t1_ms} n={n} "
+            f"t0={pressure_t0_ms} t1={t1_ms} n={n} "
-                f"min={pressure_min} max={pressure_max} avg={avg:.1f} avg_abs={avg_abs:.3f} "
+            f"min={pressure_min} max={pressure_max} avg={avg:.1f} avg_abs={avg_abs:.3f} "
-                f"values={','.join(map(str, pressure_buf))}"
+            f"values={','.join(map(str, pressure_buf))}"
-                f" convert value (kpa): {(max(pressure_buf, key=lambda x: x[1])[1] - last_avg_abs) / (5 - 2.5) * config.AIR_PRESSURE_HARDWARE_MAX:.1f}"
+            f" convert value (kpa): {(max(pressure_buf, key=lambda x: x[1])[1] - last_avg_abs) / (5 - 2.5) * config.AIR_PRESSURE_HARDWARE_MAX:.1f}"
-            )
+        )
-            if logger:
+        if logger:
-                logger.debug(line)
+            logger.debug(line)
-            else:
+        else:
-                print(line)
+            print(line)
-            last_avg_abs = avg_abs
-        # 无论是否记录日志，都必须清空 buffer，否则内存泄漏
         pressure_buf = []
         pressure_sum = 0
         pressure_abs_sum = 0
         pressure_min = 4095
         pressure_max = 0
         pressure_t0_ms = None
+        last_avg_abs = avg_abs
 
     # 主循环：检测扳机触发 → 拍照 → 分析 → 上报
     while not app.need_exit():
@@ -401,7 +402,14 @@ def cmd_str():
             else:
                 if config.SHOW_CAMERA_PHOTO_WHILE_SHOOTING:
                     try:
-                        camera_manager.show(camera_manager.read_frame())
+                        frame = camera_manager.read_frame()
+                        laser_manager.remote_detect_tick(frame)
+                        if (
+                            laser_manager.remote_detect_active
+                            and getattr(config, "LASER_REMOTE_DETECT_DRAW_PREVIEW", False)
+                        ):
+                            frame = laser_manager.overlay_remote_detect_preview(frame)
+                        camera_manager.show(frame)
                     except Exception as e:
                         logger = logger_manager.logger
                         if logger:

model_270820.mud

@@ -1,13 +0,0 @@
-
-[basic]
-type = cvimodel
-model = model_270820.cvimodel
-
-[extra]
-model_type = yolov5
-input_type = rgb
-mean = 0, 0, 0
-scale = 0.00392156862745098, 0.00392156862745098, 0.00392156862745098
-anchors = 10, 13, 16, 30, 33, 23, 30, 61, 62, 45, 59, 119, 116, 90, 156, 198, 373, 326
-labels = triangle
-

ota_curl.sh

@@ -1,57 +0,0 @@
-#!/bin/sh
-# OTA 更新脚本 - 使用 curl 断点下载
-# 用法: sh ota_curl.sh <下载URL>
-# 示例: sh ota_curl.sh http://example.com/maix-t11-v2.15.1.zip
-
-set -e
-
-APP_DIR="/maixapp/apps/t11"
-BACKUP_BASE="$APP_DIR/backups"
-TMP_DIR="/tmp/ota_curl"
-PENDING_FILE="$APP_DIR/ota_pending.json"
-
-if [ $# -lt 1 ]; then
-    echo "用法: $0 <下载URL>"
-    exit 1
-fi
-
-OTA_URL="$1"
-FILENAME=$(basename "$OTA_URL" | sed 's/?.*//')
-[ -z "$FILENAME" ] && FILENAME="update.zip"
-
-mkdir -p "$TMP_DIR" "$BACKUP_BASE"
-
-# 1. 断点下载
-echo "[OTA] 开始下载: $OTA_URL"
-echo "[OTA] 保存到: $TMP_DIR/$FILENAME"
-curl -C - -L --retry 3 --retry-delay 5 -o "$TMP_DIR/$FILENAME" "$OTA_URL"
-echo "[OTA] 下载完成"
-
-# 2. 备份当前目录
-TIMESTAMP=$(date +%Y%m%d_%H%M%S 2>/dev/null || echo "00000000_000000")
-BACKUP_DIR="$BACKUP_BASE/backup_$TIMESTAMP"
-mkdir -p "$BACKUP_DIR"
-echo "[OTA] 备份到: $BACKUP_DIR"
-for f in "$APP_DIR"/*.py "$APP_DIR"/*.json "$APP_DIR"/*.xml "$APP_DIR"/*.yaml "$APP_DIR"/*.pem "$APP_DIR"/*.mud "$APP_DIR"/*.so "$APP_DIR"/S99archery; do
-    [ -f "$f" ] && cp "$f" "$BACKUP_DIR/"
-done
-echo "[OTA] 备份完成"
-
-# 3. 解压并替换文件
-echo "[OTA] 开始更新..."
-if echo "$FILENAME" | grep -qi '\.zip$'; then
-    unzip -q -o "$TMP_DIR/$FILENAME" -d "$APP_DIR/"
-else
-    cp "$TMP_DIR/$FILENAME" "$APP_DIR/"
-fi
-sync
-
-# 4. 写入 pending 文件（用于崩溃恢复）
-echo '{"ts":0,"url":"'"$OTA_URL"'","backup_dir":"'"$BACKUP_DIR"'","restart_count":0,"max_restarts":3}' > "$PENDING_FILE"
-sync
-
-echo "[OTA] 更新完成，准备重启..."
-
-# 5. 重启
-sleep 1
-reboot

shoot_manager.py

@@ -8,7 +8,7 @@ from laser_manager import laser_manager
 from logger_manager import logger_manager
 from network import network_manager
 from triangle_target import load_camera_from_xml, load_triangle_positions, try_triangle_scoring
-from vision import estimate_distance, detect_circle_v3, enqueue_save_shot
+from vision import estimate_distance, detect_circle_v3, enqueue_save_shot, enqueue_save_raw_shot
 from maix import image, time
 
 # 缓存相机标定与三角形位置，避免每次射箭重复读磁盘
@@ -58,6 +58,7 @@ def analyze_shot(frame, laser_point=None):
     # ── Step 1: 确定激光点 ────────────────────────────────────────────────────
     laser_point_method = None
     distance_m_first = None
+    best_radius1_temp = None
 
     if config.HARDCODE_LASER_POINT:
         laser_point = laser_manager.laser_point
@@ -102,9 +103,22 @@ def analyze_shot(frame, laser_point=None):
         r_img, center, radius, method, best_radius1, ellipse_params = cdata
         dx, dy = None, None
         d_m = distance_m_first
+        tri_h = None
         if center and radius:
             dx, dy = laser_manager.compute_laser_position(center, (x, y), radius, method)
             d_m = estimate_distance(best_radius1) if best_radius1 else distance_m_first
+            try:
+                import numpy as _np
+                px_per_cm = float(radius) / 10.0
+                if px_per_cm > 1e-6:
+                    cxp, cyp = float(center[0]), float(center[1])
+                    tri_h = _np.array([
+                        [1.0 / px_per_cm, 0.0, -cxp / px_per_cm],
+                        [0.0, 1.0 / px_per_cm, -cyp / px_per_cm],
+                        [0.0, 0.0, 1.0],
+                    ], dtype=float)
+            except Exception:
+                tri_h = None
         out = {
             "success": True,
             "result_img": r_img,
@@ -114,6 +128,7 @@ def analyze_shot(frame, laser_point=None):
             "laser_point": laser_point, "laser_point_method": laser_point_method,
             "offset_method": "yellow_ellipse" if ellipse_params else "yellow_circle",
             "distance_method": "yellow_radius",
+            "tri_homography": tri_h,
         }
         if yolo_roi_xyxy is not None:
             out["yolo_roi_xyxy"] = yolo_roi_xyxy
@@ -129,8 +144,10 @@ def analyze_shot(frame, laser_point=None):
     roi_xyxy = None
     yolo_ring_ms = 0.0
     yolo_black_ms = 0.0
+    _timing_on = bool(getattr(config, "ARCHERY_TIMING_ENABLE", True))
+    _sample_on = bool(getattr(config, "TRIANGLE_SAMPLE_ENABLE", False))
     if getattr(config, "TRIANGLE_YOLO_ROI_ENABLE", False):
-        _t_yolo_ring = time_std.perf_counter()
+        _t_yolo_ring = time_std.perf_counter() if _timing_on else None
         try:
             from target_roi_yolo import try_get_triangle_roi_from_yolo
             roi_xyxy = try_get_triangle_roi_from_yolo(
@@ -140,7 +157,8 @@ def analyze_shot(frame, laser_point=None):
             if logger:
                 logger.warning(f"[YOLO-ROI] {e}")
         finally:
-            yolo_ring_ms = (time_std.perf_counter() - _t_yolo_ring) * 1000.0
+            if _timing_on and _t_yolo_ring is not None:
+                yolo_ring_ms = (time_std.perf_counter() - _t_yolo_ring) * 1000.0
 
     _loc_mode = str(
         getattr(config, "TRIANGLE_BLACK_TRIANGLE_LOCATE_MODE", "yolo")
@@ -155,7 +173,7 @@ def analyze_shot(frame, laser_point=None):
         and roi_xyxy is not None
     )
     if _run_stage2_black_yolo:
-        _t_yolo_black = time_std.perf_counter()
+        _t_yolo_black = time_std.perf_counter() if _timing_on else None
         try:
             from target_roi_yolo import try_black_triangle_boxes_work
 
@@ -166,7 +184,8 @@ def analyze_shot(frame, laser_point=None):
             if logger:
                 logger.warning(f"[YOLO-BLACK] {e}")
         finally:
-            yolo_black_ms = (time_std.perf_counter() - _t_yolo_black) * 1000.0
+            if _timing_on and _t_yolo_black is not None:
+                yolo_black_ms = (time_std.perf_counter() - _t_yolo_black) * 1000.0
     elif (
         logger
         and _loc_mode == "traditional"
@@ -184,7 +203,7 @@ def analyze_shot(frame, laser_point=None):
         try:
             logger.info(f"[TRI] begin {datetime.now()}")
             logger.info(f"[TRI] K: {K}, dist: {dist_coef}, pos: {pos},  {datetime.now()}")
-            _t_wall_try = time_std.perf_counter()
+            _t_wall_try = time_std.perf_counter() if _timing_on else None
             tri = try_triangle_scoring(
                 img_cv, (x, y), pos, K, dist_coef,
                 size_range=getattr(config, "TRIANGLE_SIZE_RANGE", (8, 500)),
@@ -193,8 +212,8 @@ def analyze_shot(frame, laser_point=None):
                 yolo_ring_ms=yolo_ring_ms,
                 yolo_black_ms=yolo_black_ms,
             )
-            _wall_try_ms = (time_std.perf_counter() - _t_wall_try) * 1000.0
+            _wall_try_ms = (time_std.perf_counter() - _t_wall_try) * 1000.0 if _timing_on else 0.0
-            if logger and bool(getattr(config, "TRIANGLE_LOG_E2E_TIMING", True)):
+            if logger and bool(getattr(config, "TRIANGLE_LOG_E2E_TIMING", True)) and _timing_on:
                 _e2e = float(yolo_ring_ms) + float(yolo_black_ms) + float(_wall_try_ms)
                 logger.info(
                     f"[TRI] timing_e2e_triangle_ms={_e2e:.1f} "
@@ -280,6 +299,16 @@ def analyze_shot(frame, laser_point=None):
             "tri_markers_completed": tri.get("markers_completed", []),
             "tri_homography": tri.get("homography"),
         }
+        try:
+            import numpy as _np
+            _H = tri.get("homography")
+            if _H is not None and _np.all(_np.isfinite(_H)):
+                _H_inv = _np.linalg.inv(_H)
+                _pt = _np.array([[[0.0, 0.0]]], dtype=_np.float32)
+                _center_pt = cv2.perspectiveTransform(_pt, _H_inv)[0][0]
+                out["tri_center_px"] = [float(_center_pt[0]), float(_center_pt[1])]
+        except Exception:
+            pass
         if yolo_roi_xyxy is not None:
             out["yolo_roi_xyxy"] = yolo_roi_xyxy
         return out
@@ -318,11 +347,22 @@ def process_shot(adc_val):
     :return: None
     """
     logger = logger_manager.logger
+    _timing_on = bool(getattr(config, "ARCHERY_TIMING_ENABLE", True))
 
     try:
         network_manager.safe_enqueue({"shoot_event": "start"}, msg_type=2, high=True)
         frame = camera_manager.read_frame()
 
+        from shot_id_generator import shot_id_generator
+        shot_id = shot_id_generator.generate_id()
+
+        if getattr(config, "SAVE_RAW_SHOT_IMAGE_ENABLED", False):
+            enqueue_save_raw_shot(
+                frame,
+                shot_id=shot_id,
+                photo_dir=config.PHOTO_DIR if config.SAVE_IMAGE_ENABLED else None,
+            )
+
         # 调用算法分析
         analysis_result = analyze_shot(frame)
 
@@ -356,6 +396,107 @@ def process_shot(adc_val):
         )
         x, y = laser_point
 
+        # 物方采样调试（config.TRIANGLE_SAMPLE_ENABLE）：靶心为原点，取两个对称点判断黑白来区分 40/20 标靶
+        # 逻辑：若两个采样点 RGB 均 < 阈值 → 全黑 → 40cm 标靶；否则 → 20cm 标靶
+        sample_target_type = None
+        _t_sample = time_std.perf_counter() if _timing_on else None
+        _t_sample_ms = 0.0
+        sample_points = []
+        sample_patch_half = 2
+        if bool(getattr(config, "TRIANGLE_SAMPLE_ENABLE", False)):
+            sample_obj_radius_cm = float(getattr(config, "TRIANGLE_SAMPLE_RADIUS_CM", 15.0))
+            sample_obj_angles_deg = (0, 180)  # 只取两个对称点：+X 和 -X
+            sample_patch_half = int(getattr(config, "TRIANGLE_SAMPLE_PATCH_HALF_PX", 2))
+            sample_black_thresh = float(getattr(config, "TRIANGLE_SAMPLE_BLACK_THRESH", 30.0))
+            try:
+                import math as _math
+                import numpy as _np
+                import cv2 as _cv2
+
+                if tri_homography is not None:
+                    _H_inv = _np.linalg.inv(tri_homography)
+                    for _ang in sample_obj_angles_deg:
+                        _rad = _math.radians(float(_ang))
+                        _pt_obj = _np.array([
+                            [[sample_obj_radius_cm * _math.cos(_rad), sample_obj_radius_cm * _math.sin(_rad)]]
+                        ], dtype=_np.float32)
+                        _pt_img = _cv2.perspectiveTransform(_pt_obj, _H_inv)[0][0]
+                        _px, _py = float(_pt_img[0]), float(_pt_img[1])
+                        sample_points.append({
+                            "angle_deg": float(_ang),
+                            "obj_cm": (float(sample_obj_radius_cm * _math.cos(_rad)), float(sample_obj_radius_cm * _math.sin(_rad))),
+                            "img_px": (int(round(_px)), int(round(_py))),
+                        })
+                elif center and radius:
+                    _px_per_cm = float(radius) / 10.0
+                    for _ang in sample_obj_angles_deg:
+                        _rad = _math.radians(float(_ang))
+                        _px = float(center[0]) + sample_obj_radius_cm * _math.cos(_rad) * _px_per_cm
+                        _py = float(center[1]) + sample_obj_radius_cm * _math.sin(_rad) * _px_per_cm
+                        sample_points.append({
+                            "angle_deg": float(_ang),
+                            "obj_cm": (float(sample_obj_radius_cm * _math.cos(_rad)), float(sample_obj_radius_cm * _math.sin(_rad))),
+                            "img_px": (int(round(_px)), int(round(_py))),
+                        })
+
+                # 取样后立即读像素并判断黑白：三角成功用 H_inv；三角失败但圆心成功用 center/radius 近似物方半径
+                _all_black = False
+                _sample_infos = []
+                if sample_points:
+                    _img_cv_for_sample = image.image2cv(result_img, False, False)
+                    _all_black = True
+                    for _sp in sample_points:
+                        _sx, _sy = _sp["img_px"]
+                        _hh = max(1, sample_patch_half)
+                        _patch = []
+                        for _yy in range(_sy - _hh, _sy + _hh + 1):
+                            if _yy < 0 or _yy >= _img_cv_for_sample.shape[0]:
+                                continue
+                            for _xx in range(_sx - _hh, _sx + _hh + 1):
+                                if _xx < 0 or _xx >= _img_cv_for_sample.shape[1]:
+                                    continue
+                                _patch.append(_img_cv_for_sample[_yy, _xx].astype(float))
+                        if _patch:
+                            _mean_rgb = _np.mean(_patch, axis=0)
+                            _is_black = bool(_mean_rgb[0] < sample_black_thresh
+                                           and _mean_rgb[1] < sample_black_thresh
+                                           and _mean_rgb[2] < sample_black_thresh)
+                            if not _is_black:
+                                _all_black = False
+                            _sample_infos.append(
+                                f"{int(_sp['angle_deg'])}°@{_sx},{_sy} rgb=({int(_mean_rgb[0])},{int(_mean_rgb[1])},{int(_mean_rgb[2])})"
+                            )
+                    sample_target_type = "40cm_black" if _all_black else "20cm"
+                if _sample_infos:
+                    logger.info("[采样] " + " | ".join(_sample_infos) + f" → {sample_target_type}")
+            except Exception as _e_sample:
+                sample_points = []
+                if logger:
+                    logger.warning(f"[采样] 标靶类型判断失败: {_e_sample}")
+        if _timing_on and _t_sample is not None:
+            _t_sample_ms = (time_std.perf_counter() - _t_sample) * 1000.0
+
+        # 采样提前完成后，先确定靶型对应的物理半径，供后续距离/偏移/上报使用。
+        # 40cm_black 表示直径40cm，半径20cm；20cm 表示直径20cm，半径10cm。
+        target_radius_cm = 20.0 if sample_target_type == "40cm_black" else (10.0 if sample_target_type == "20cm" else 20.0)
+        target_type_value = 40 if sample_target_type == "40cm_black" else (20 if sample_target_type == "20cm" else None)
+
+        # 圆心分支原算法默认按40cm靶半径20cm换算；若采样判定为20cm靶，在上报前修正距离和偏移。
+        # 三角分支使用 triangle_positions.json 的物方坐标，不在这里二次缩放，避免影响三角单应性结果。
+        if sample_target_type == "20cm" and center and radius and not tri_markers:
+            try:
+                distance_m = (target_radius_cm * config.FOCAL_LENGTH_PIX) / float(radius) / 100.0
+                _scale = target_radius_cm / 20.0
+                if dx is not None:
+                    dx = float(dx) * _scale
+                if dy is not None:
+                    dy = float(dy) * _scale
+                if logger:
+                    logger.info(f"[采样] 20cm靶修正圆心测距/偏移: distance={distance_m:.2f}m scale={_scale:.2f}")
+            except Exception as _e_fix:
+                if logger:
+                    logger.warning(f"[采样] 20cm靶修正失败: {_e_fix}")
+
         # 三角形路径成功时 center/radius 为空是正常的；此时用 triangle 方法名用于保存文件名与上报字段 m
         if (not method) and tri_markers:
             method = "triangle_homography"
@@ -366,10 +507,6 @@ def process_shot(adc_val):
         if dx is None and dy is None and logger:
             logger.warning("[MAIN] 未检测到偏移量（三角形与圆形均失败），但会保存图像")
 
-        # 生成射箭ID
-        from shot_id_generator import shot_id_generator
-        shot_id = shot_id_generator.generate_id()
-
         if logger:
             logger.info(f"[MAIN] 射箭ID: {shot_id}")
 
@@ -386,7 +523,7 @@ def process_shot(adc_val):
             "shot_id": shot_id,
             "x": srv_x,
             "y": srv_y,
-            "r": 20.0,  # 保留字段（服务端当前忽略，物理外环半径 cm）
+            "r": target_radius_cm,  # 物理靶半径 cm：40cm靶=20，20cm靶=10
             "d": round((distance_m or 0.0) * 100),
             "d_laser": round((laser_distance_m or 0.0) * 100),
             "d_laser_quality": laser_signal_quality,
@@ -397,6 +534,7 @@ def process_shot(adc_val):
             "target_y": float(y),
             "offset_method": offset_method,
             "distance_method": distance_method,
+            "target_type": target_type_value,
         }
 
         if ellipse_params:
@@ -471,6 +609,11 @@ def process_shot(adc_val):
                     except Exception:
                         pass
 
+                # 物方采样标靶类型判断耗时（合并在上面采样块内，单独统计）
+                if _timing_on and bool(getattr(config, "TRIANGLE_SAMPLE_ENABLE", False)) and sample_target_type is not None:
+                    logger.info(f"[采样] 标靶类型: {sample_target_type}  耗时: {_t_sample_ms:.2f}ms")
+
+
                 # 叠加信息：落点-圆心距离 / 相机-靶距离等
                 try:
                     import math as _math

test/test_algo_preview_live.py

@@ -0,0 +1,403 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+实时摄像头预览：叠加与射箭存图相同的算法标注（YOLO ROI、三角/圆心、激光十字等），默认不写盘。
+
+在 MaixCAM 上从项目根目录运行：
+  python3 test/test_algo_preview_live.py
+  python3 test/test_algo_preview_live.py --interval 1.5
+  python3 test/test_algo_preview_live.py --every-frame
+
+说明：
+  - 完整算法走 shoot_manager.analyze_shot（与 process_shot 一致，含 YOLO + 三角/圆心）。
+  - 画面标注对齐 process_shot 存图前绘制 + vision._draw_yolo_roi_on_rgb_numpy / 圆心存图线。
+  - 预览模式会关闭 Stage2 裁切 JPEG 落盘，避免写满 /root/phot。
+"""
+
+from __future__ import annotations
+
+import argparse
+import math
+import os
+import sys
+import threading
+import time
+
+_ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+if _ROOT not in sys.path:
+    sys.path.insert(0, _ROOT)
+
+import cv2
+import numpy as np
+from maix import image, time as maix_time
+
+import config
+from camera_manager import camera_manager
+from laser_manager import laser_manager
+from shoot_manager import analyze_shot, preload_triangle_calib
+from target_roi_yolo import preload_yolo_detector
+from vision import _draw_yolo_roi_on_rgb_numpy
+
+
+def _copy_maix_frame(frame):
+    """相机下一帧可能复用缓冲区，异步分析前先复制。"""
+    img_cv = image.image2cv(frame, False, False)
+    return image.cv2image(np.ascontiguousarray(img_cv), False, False)
+
+
+def _patch_preview_config():
+    """预览不写调试 JPEG，避免刷屏占存储。"""
+    config.TRIANGLE_BLACK_YOLO_SAVE_ROI_CROP = False
+    config.TRIANGLE_SAVE_DEBUG_IMAGE = False
+
+
+def _annotate_like_saved_shot(analysis: dict):
+    """
+    将 analyze_shot 结果绘制成与 process_shot -> enqueue_save_shot 存盘前一致的 Maix 图。
+    """
+    result_img = analysis.get("result_img")
+    if result_img is None:
+        return None
+
+    center = analysis.get("center")
+    radius = analysis.get("radius")
+    method = analysis.get("method")
+    ellipse_params = analysis.get("ellipse_params")
+    laser_point = analysis.get("laser_point")
+    dx = analysis.get("dx")
+    dy = analysis.get("dy")
+    distance_m = analysis.get("distance_m")
+    offset_method = analysis.get("offset_method", "")
+    distance_method = analysis.get("distance_method", "")
+    tri_markers = analysis.get("tri_markers") or []
+    tri_markers_completed = analysis.get("tri_markers_completed") or []
+    tri_homography = analysis.get("tri_homography")
+    yolo_roi_xyxy = analysis.get("yolo_roi_xyxy")
+
+    if laser_point is None:
+        return result_img
+
+    x, y = laser_point
+    draw_yolo_roi = (
+        yolo_roi_xyxy is not None
+        and getattr(config, "TRIANGLE_YOLO_DRAW_ROI_ON_SHOT", True)
+    )
+
+    if tri_markers:
+        img_cv = image.image2cv(result_img, False, False).copy()
+
+        if draw_yolo_roi:
+            _draw_yolo_roi_on_rgb_numpy(img_cv, yolo_roi_xyxy)
+
+        for m in tri_markers:
+            corners = np.array(m["corners"], dtype=np.int32)
+            cv2.polylines(img_cv, [corners], True, (0, 255, 0), 2)
+            cx, cy = int(m["center"][0]), int(m["center"][1])
+            cv2.circle(img_cv, (cx, cy), 4, (0, 0, 255), -1)
+            cv2.putText(
+                img_cv,
+                f"T{m['id']}",
+                (cx - 18, cy - 12),
+                cv2.FONT_HERSHEY_SIMPLEX,
+                0.55,
+                (0, 255, 0),
+                1,
+            )
+
+        for m in tri_markers_completed:
+            if not m.get("is_virtual"):
+                continue
+            cx, cy = int(m["center"][0]), int(m["center"][1])
+            cv2.circle(img_cv, (cx, cy), 6, (255, 0, 255), 2)
+            cv2.putText(
+                img_cv,
+                f"VT{m['id']}",
+                (cx - 22, cy - 12),
+                cv2.FONT_HERSHEY_SIMPLEX,
+                0.55,
+                (255, 0, 255),
+                1,
+            )
+
+        if tri_homography is not None:
+            try:
+                H_inv = np.linalg.inv(tri_homography)
+                c_img = cv2.perspectiveTransform(
+                    np.array([[[0.0, 0.0]]], dtype=np.float32), H_inv
+                )[0][0]
+                ocx, ocy = int(c_img[0]), int(c_img[1])
+                cv2.circle(img_cv, (ocx, ocy), 5, (0, 0, 255), -1)
+                cv2.circle(img_cv, (ocx, ocy), 9, (0, 0, 255), 1)
+            except Exception:
+                pass
+
+        lines = []
+        if dx is not None and dy is not None:
+            r_cm = math.hypot(float(dx), float(dy))
+            lines.append(f"offset=({float(dx):.2f},{float(dy):.2f})cm |r|={r_cm:.2f}cm")
+        if distance_m is not None:
+            lines.append(f"cam_dist={float(distance_m):.2f}m ({distance_method})")
+        if method:
+            lines.append(f"method={method} ({offset_method})")
+        y0 = 22
+        for i, t in enumerate(lines):
+            cv2.putText(
+                img_cv,
+                t,
+                (10, y0 + i * 18),
+                cv2.FONT_HERSHEY_SIMPLEX,
+                0.5,
+                (0, 255, 0),
+                1,
+            )
+
+        out = image.cv2image(img_cv, False, False)
+    else:
+        img_cv = image.image2cv(result_img, False, False).copy()
+        if draw_yolo_roi:
+            _draw_yolo_roi_on_rgb_numpy(img_cv, yolo_roi_xyxy)
+
+        if center and radius:
+            cx, cy = center
+            if ellipse_params:
+                (ell_center, (width, height), angle) = ellipse_params
+                cx_ell, cy_ell = int(ell_center[0]), int(ell_center[1])
+                cv2.ellipse(
+                    img_cv,
+                    (cx_ell, cy_ell),
+                    (int(width / 2), int(height / 2)),
+                    angle,
+                    0,
+                    360,
+                    (0, 255, 0),
+                    2,
+                )
+                cv2.circle(img_cv, (cx_ell, cy_ell), 3, (255, 0, 0), -1)
+                minor_length = min(width, height) / 2
+                minor_angle = angle + 90 if width >= height else angle
+                minor_angle_rad = math.radians(minor_angle)
+                dx_minor = minor_length * math.cos(minor_angle_rad)
+                dy_minor = minor_length * math.sin(minor_angle_rad)
+                pt1 = (int(cx_ell - dx_minor), int(cy_ell - dy_minor))
+                pt2 = (int(cx_ell + dx_minor), int(cy_ell + dy_minor))
+                cv2.line(img_cv, pt1, pt2, (0, 0, 255), 2)
+            else:
+                cv2.circle(img_cv, (int(cx), int(cy)), int(radius), (0, 0, 255), 2)
+                cv2.circle(img_cv, (int(cx), int(cy)), 2, (0, 0, 255), -1)
+            cv2.line(img_cv, (int(x), int(y)), (int(cx), int(cy)), (255, 255, 0), 1)
+
+        lines = []
+        if dx is not None and dy is not None:
+            lines.append(f"offset=({float(dx):.2f},{float(dy):.2f})cm")
+        if distance_m is not None:
+            lines.append(f"dist={float(distance_m):.2f}m ({distance_method})")
+        if method:
+            lines.append(f"method={method}")
+        for i, t in enumerate(lines):
+            cv2.putText(
+                img_cv,
+                t,
+                (10, 22 + i * 18),
+                cv2.FONT_HERSHEY_SIMPLEX,
+                0.5,
+                (0, 255, 0),
+                1,
+            )
+        out = image.cv2image(img_cv, False, False)
+
+    lc = image.Color(config.LASER_COLOR[0], config.LASER_COLOR[1], config.LASER_COLOR[2])
+    out.draw_line(
+        int(x - config.LASER_LENGTH),
+        int(y),
+        int(x + config.LASER_LENGTH),
+        int(y),
+        lc,
+        config.LASER_THICKNESS,
+    )
+    out.draw_line(
+        int(x),
+        int(y - config.LASER_LENGTH),
+        int(x),
+        int(y + config.LASER_LENGTH),
+        lc,
+        config.LASER_THICKNESS,
+    )
+    out.draw_circle(int(x), int(y), 1, lc, config.LASER_THICKNESS)
+    return out
+
+
+class _AlgoWorker:
+    def __init__(self):
+        self._lock = threading.Lock()
+        self._busy = False
+        self._latest_preview = None
+        self._latest_meta = ""
+        self._last_ms = 0.0
+
+    @property
+    def busy(self):
+        with self._lock:
+            return self._busy
+
+    @property
+    def last_ms(self):
+        with self._lock:
+            return self._last_ms
+
+    def get_preview(self):
+        with self._lock:
+            return self._latest_preview, self._latest_meta
+
+    def run_async(self, frame):
+        with self._lock:
+            if self._busy:
+                return False
+            self._busy = True
+
+        def _job():
+            t0 = time.perf_counter()
+            meta = ""
+            preview = None
+            try:
+                analysis = analyze_shot(frame)
+                if not analysis.get("success"):
+                    reason = analysis.get("reason", "unknown")
+                    meta = f"fail:{reason}"
+                else:
+                    preview = _annotate_like_saved_shot(analysis)
+                    dx, dy = analysis.get("dx"), analysis.get("dy")
+                    method = analysis.get("method") or "?"
+                    if dx is not None and dy is not None:
+                        meta = f"ok {method} ({dx:.2f},{dy:.2f})cm"
+                    else:
+                        meta = f"ok {method} no_offset"
+            except Exception as e:
+                meta = f"err:{e}"
+            elapsed = (time.perf_counter() - t0) * 1000.0
+            with self._lock:
+                self._latest_preview = preview
+                self._latest_meta = f"{meta} {elapsed:.0f}ms"
+                self._last_ms = elapsed
+                self._busy = False
+
+        threading.Thread(target=_job, daemon=True).start()
+        return True
+
+
+def _draw_status(frame, lines, color=None):
+    if color is None:
+        color = image.COLOR_YELLOW
+    y = 4
+    for line in lines:
+        frame.draw_string(4, y, line, color=color)
+        y += 16
+
+
+def _save_preview_jpeg(maix_img, out_dir):
+    os.makedirs(out_dir, exist_ok=True)
+    fn = os.path.join(out_dir, f"preview_{int(time.time() * 1000)}.jpg")
+    maix_img.save(fn)
+    return fn
+
+
+def main():
+    parser = argparse.ArgumentParser(description="实时预览射箭算法存图效果")
+    parser.add_argument(
+        "--interval",
+        type=float,
+        default=2.0,
+        help="两次完整 analyze_shot 的最小间隔（秒）；--every-frame 时忽略",
+    )
+    parser.add_argument(
+        "--every-frame",
+        action="store_true",
+        help="每帧都触发算法（很慢，仅调试用）",
+    )
+    parser.add_argument(
+        "--width",
+        type=int,
+        default=getattr(config, "CAMERA_WIDTH", 640),
+    )
+    parser.add_argument(
+        "--height",
+        type=int,
+        default=getattr(config, "CAMERA_HEIGHT", 480),
+    )
+    parser.add_argument(
+        "--save-dir",
+        default=config.PHOTO_DIR,
+        help="按板子按键无；用 --save-every N 每 N 次成功分析存一张",
+    )
+    parser.add_argument(
+        "--save-every",
+        type=int,
+        default=0,
+        help="每成功分析 N 次自动存一张到 --save-dir（0=不自动存）",
+    )
+    args = parser.parse_args()
+
+    _patch_preview_config()
+    print("[INFO] 预览模式：已关闭 TRIANGLE_BLACK_YOLO_SAVE_ROI_CROP / TRIANGLE_SAVE_DEBUG_IMAGE")
+
+    laser_manager.load_laser_point()
+    preload_triangle_calib()
+    if getattr(config, "TRIANGLE_YOLO_PRELOAD_ON_BOOT", False) or getattr(
+        config, "TRIANGLE_BLACK_YOLO_PRELOAD_ON_BOOT", False
+    ):
+        print("[INFO] 预加载 YOLO …")
+        preload_yolo_detector()
+
+    camera_manager.init_camera(args.width, args.height)
+    camera_manager.init_display()
+    worker = _AlgoWorker()
+
+    interval_s = 0.0 if args.every_frame else max(0.3, float(args.interval))
+    last_trigger = 0.0
+    ok_count = 0
+    frame_idx = 0
+
+    print(
+        f"[INFO] 摄像头 {args.width}x{args.height} "
+        f"interval={'每帧' if args.every_frame else f'{interval_s}s'}"
+    )
+    print("[INFO] 退出：Ctrl+C")
+
+    try:
+        while True:
+            frame = camera_manager.read_frame()
+            frame_idx += 1
+            now = time.perf_counter()
+
+            due = args.every_frame or (now - last_trigger >= interval_s)
+            if due and not worker.busy:
+                last_trigger = now
+                worker.run_async(_copy_maix_frame(frame))
+
+            preview, meta = worker.get_preview()
+            if preview is not None:
+                show_img = preview
+                status = [f"#{frame_idx}", meta]
+                if args.save_every > 0 and meta.startswith("ok"):
+                    ok_count += 1
+                    if ok_count % args.save_every == 0:
+                        try:
+                            fn = _save_preview_jpeg(preview, args.save_dir)
+                            status.append(f"saved:{fn}")
+                        except Exception as e:
+                            status.append(f"save_err:{e}")
+            else:
+                show_img = frame
+                if worker.busy:
+                    status = [f"#{frame_idx}", "analyzing…"]
+                else:
+                    status = [f"#{frame_idx}", "waiting…"]
+
+            _draw_status(show_img, status)
+            camera_manager.show(show_img)
+            maix_time.sleep_ms(1)
+    except KeyboardInterrupt:
+        print("[INFO] 已退出")
+
+
+if __name__ == "__main__":
+    main()

test/test_decect.py

@@ -1,330 +0,0 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-离线测试脚本：直接复用 detect_circle 逻辑进行测试
-运行环境：MaixPy (Sipeed MAIX)
-"""
-import sys
-import os
-# import time
-from maix import image, time
-import cv2
-import numpy as np
-import math
-
-# ==================== 全局配置 (与 test_main.py 保持一致) ====================
-REAL_RADIUS_CM = 20  # 靶心实际半径（厘米）
-
-def detect_circle_v3(frame, laser_point=None, img_cv=None):
-    """检测图像中的靶心（优先清晰轮廓，其次黄色区域）- 返回椭圆参数版本
-    增加红色圆圈检测，验证黄色圆圈是否为真正的靶心
-    如果提供 laser_point，会选择最接近激光点的目标
-    优化：
-      1. 缩图到 MAX_DET_DIM 后再做 HSV/形态学，最长边 640->320 可获得 ~4x 加速
-      2. 红色掩码在黄色轮廓循环外只计算一次，避免 N 次重复计算
-      3. img_cv 可由外部传入（与其他线程共享转换结果），为 None 时自动转换
-    Args:
-        frame: 图像帧（img_cv 为 None 时使用）
-        laser_point: 激光点坐标 (x, y)，用于多目标场景下的目标选择
-        img_cv: 已转换的 numpy BGR/RGB 图像；不为 None 时跳过 image2cv 转换
-    Returns:
-        (result_img, best_center, best_radius, method, best_radius1, ellipse_params)
-    """
-    if img_cv is None:
-        img_cv = image.image2cv(frame, False, False)
-    from datetime import datetime
-    print(f"[detect_circle_v3] begin {datetime.now()}")
-    # -- 1. 缩图加速（与三角形路径保持一致）
-    h_orig, w_orig = img_cv.shape[:2]
-    MAX_DET_DIM = 480
-    long_side = max(h_orig, w_orig)
-    if long_side > MAX_DET_DIM:
-        det_scale = MAX_DET_DIM / long_side
-        img_det = cv2.resize(img_cv, (int(w_orig * det_scale), int(h_orig * det_scale)),
-                             interpolation=cv2.INTER_LINEAR)
-        inv_scale = 1.0 / det_scale  # 检测坐标 -> 原始坐标的倍率
-    else:
-        img_det = img_cv
-        inv_scale = 1.0
-
-    # 激光点映射到检测分辨率
-    lp_det = None
-    if laser_point is not None:
-        lp_det = (laser_point[0] / inv_scale, laser_point[1] / inv_scale)
-    best_center = best_radius = best_radius1 = method = None
-    ellipse_params = None
-
-    print(f"[detect_circle_v3] step 1 fin {datetime.now()}")
-
-    # -- 2. HSV + 黄色掩码
-    hsv = cv2.cvtColor(img_det, cv2.COLOR_RGB2HSV)
-    h, s, v = cv2.split(hsv)
-    s = np.clip(s * 1.1, 0, 255).astype(np.uint8)
-    hsv = cv2.merge((h, s, v))
-    lower_yellow = np.array([7, 80, 0])
-    upper_yellow = np.array([32, 255, 255])
-    mask_yellow = cv2.inRange(hsv, lower_yellow, upper_yellow)
-    kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
-    mask_yellow = cv2.morphologyEx(mask_yellow, cv2.MORPH_CLOSE, kernel)
-
-    print(f"[detect_circle_v3] step 2 fin {datetime.now()}")
-
-    # -- 3. 红色掩码：在循环外只算一次
-    mask_red = cv2.bitwise_or(
-        cv2.inRange(hsv, np.array([0, 50, 40]), np.array([10, 255, 255])),
-        cv2.inRange(hsv, np.array([170, 50, 40]), np.array([180, 255, 255])),
-    )
-    kernel_red = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
-    mask_red = cv2.morphologyEx(mask_red, cv2.MORPH_CLOSE, kernel_red)
-    contours_red, _ = cv2.findContours(mask_red, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-    # 预先把红色轮廓筛选成 (center, radius) 列表，后续直接查表
-    red_candidates = []
-    for cnt_r in contours_red:
-        ar = cv2.contourArea(cnt_r)
-        if ar <= 10:
-            continue
-        pr = cv2.arcLength(cnt_r, True)
-        if pr <= 0 or (4 * np.pi * ar) / (pr * pr) <= 0.3:
-            continue
-        if len(cnt_r) >= 5:
-            (xr, yr), (wr, hr), _ = cv2.fitEllipse(cnt_r)
-            red_candidates.append({"center": (int(xr), int(yr)), "radius": int(min(wr, hr) / 2)})
-        else:
-            (xr, yr), rr = cv2.minEnclosingCircle(cnt_r)
-            red_candidates.append({"center": (int(xr), int(yr)), "radius": int(rr)})
-
-    print(f"[detect_circle_v3] step 3 fin {datetime.now()}")
-
-    # -- 4. 黄色轮廓循环（复用上面的红色候选列表）
-    contours_yellow, _ = cv2.findContours(mask_yellow, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-    valid_targets = []
-    for cnt_yellow in contours_yellow:
-        area = cv2.contourArea(cnt_yellow)
-        if area <= 15:
-            continue
-        perimeter = cv2.arcLength(cnt_yellow, True)
-        if perimeter <= 0:
-            continue
-        circularity = (4 * np.pi * area) / (perimeter * perimeter)
-        if circularity <= 0.5:
-            continue
-        print(f"[target] -> 面积:{area:.1f}, 圆度:{circularity:.2f}")
-        if len(cnt_yellow) >= 5:
-            (x, y), (width, height), angle = cv2.fitEllipse(cnt_yellow)
-            yellow_ellipse = ((x, y), (width, height), angle)
-            yellow_center = (int(x), int(y))
-            yellow_radius = int(min(width, height) / 2)
-        else:
-            (x, y), radius = cv2.minEnclosingCircle(cnt_yellow)
-            yellow_center = (int(x), int(y))
-            yellow_radius = int(radius)
-            yellow_ellipse = None
-        # 在预筛好的红色候选中匹配
-        matched = False
-        for rc in red_candidates:
-            ddx = yellow_center[0] - rc["center"][0]
-            ddy = yellow_center[1] - rc["center"][1]
-            dist_centers = math.hypot(ddx, ddy)
-            if dist_centers < yellow_radius * 1.5 and rc["radius"] > yellow_radius * 0.7:
-                print(f"[target] -> 找到匹配的红圈: 黄心({yellow_center}), "
-                                f"红心({rc['center']}), 距离:{dist_centers:.1f}, "
-                                f"黄半径:{yellow_radius}, 红半径:{rc['radius']}")
-                valid_targets.append({
-                    "center": yellow_center,
-                    "radius": yellow_radius,
-                    "ellipse": yellow_ellipse,
-                    "area": area,
-                })
-                matched = True
-                break
-        if not matched :
-            print("Debug -> 未找到匹配的红色圆圈，可能是误识别")
-
-    print(f"[detect_circle_v3] step 4 fin {datetime.now()}")
-
-    # -- 5. 选最佳目标，坐标还原到原始分辨率
-    if valid_targets:
-        if lp_det:
-            best_target = min(valid_targets,
-                              key=lambda t: (t["center"][0] - lp_det[0]) ** 2
-                                            + (t["center"][1] - lp_det[1]) ** 2)
-            method = "v3_ellipse_red_validated_laser_selected"
-        else:
-            best_target = max(valid_targets, key=lambda t: t["area"])
-            method = "v3_ellipse_red_validated"
-        bc = best_target["center"]
-        br = best_target["radius"]
-        be = best_target["ellipse"]
-        if inv_scale != 1.0:
-            best_center = (int(bc[0] * inv_scale), int(bc[1] * inv_scale))
-            best_radius = int(br * inv_scale)
-            if be is not None:
-                (ex, ey), (ew, eh), ea = be
-                be = ((ex * inv_scale, ey * inv_scale),
-                      (ew * inv_scale, eh * inv_scale), ea)
-        else:
-            best_center = bc
-            best_radius = br
-        ellipse_params = be
-        best_radius1 = best_radius * 5
-    result_img = image.cv2image(img_cv, False, False)
-    print(f"[detect_circle_v3] step 5 fin {datetime.now()}")
-    return result_img, best_center, best_radius, method, best_radius1, ellipse_params
-
-
-def run_offline_test(image_path):
-    """读取图片，检测圆，绘制结果，保存图片"""
-
-    # 1. 检查文件是否存在
-    if not os.path.exists(image_path):
-        print(f"[ERROR] 找不到图片文件: {image_path}")
-        return
-
-    # 2. 使用 maix.image 读取图片 (适配 MaixPy v4)
-    try:
-        # 使用 image.load 读取文件，返回 Image 对象
-        img = image.load(image_path)
-        print(f"[INFO] 成功读取图片: {image_path} (尺寸: {img.width()}x{img.height()})")
-    except Exception as e:
-        print(f"[ERROR] 读取图片失败: {e}")
-        print("提示：请确认 MaixPy 版本是否为 v4，且图片路径正确。")
-        return
-
-    # 3. 调用 detect_circle_v3 函数
-    print("[INFO] 正在调用 detect_circle_v3 进行检测...")
-    start_time = time.ticks_ms()
-
-    result_img, center, radius, method, radius1, ellipse_params = detect_circle_v3(img)
-
-    cost_time = time.ticks_ms() - start_time
-    print(f"[INFO] 检测完成，耗时: {cost_time}ms")
-    print(f"       结果 -> 圆心: {center}, 半径: {radius}, 方法: {method}")
-    if ellipse_params:
-        (ell_center, (width, height), angle) = ellipse_params
-        print(
-            f"       椭圆 -> 中心: ({ell_center[0]:.1f}, {ell_center[1]:.1f}), 长轴: {max(width, height):.1f}, 短轴: {min(width, height):.1f}, 角度: {angle:.1f}°")
-
-    # 4. 绘制辅助线（可选，用于调试）
-    if center and radius:
-        # 为了绘制椭圆，需要转换回 cv2 图像
-        img_cv = image.image2cv(result_img, False, False)
-
-        cx, cy = center
-
-        # 如果有椭圆参数，绘制椭圆
-        if ellipse_params:
-            (ell_center, (width, height), angle) = ellipse_params
-            cx_ell, cy_ell = int(ell_center[0]), int(ell_center[1])
-
-            # 确定长轴和短轴
-            if width >= height:
-                # width 是长轴，height 是短轴
-                axes_major = width
-                axes_minor = height
-                major_angle = angle  # 长轴角度就是 angle
-                minor_angle = angle + 90  # 短轴角度 = 长轴角度 + 90度
-            else:
-                # height 是长轴，width 是短轴
-                axes_major = height
-                axes_minor = width
-                major_angle = angle + 90  # 长轴角度 = width角度 + 90度
-                minor_angle = angle  # 短轴角度就是 angle
-
-            # 使用 OpenCV 绘制椭圆（绿色，线宽2）
-            cv2.ellipse(img_cv,
-                        (cx_ell, cy_ell),  # 中心点
-                        (int(width / 2), int(height / 2)),  # 半宽、半高
-                        angle,  # 旋转角度（OpenCV需要原始angle）
-                        0, 360,  # 起始和结束角度
-                        (0, 255, 0),  # 绿色 (RGB格式)
-                        2)  # 线宽
-
-            # 绘制椭圆中心点（红色）
-            cv2.circle(img_cv, (cx_ell, cy_ell), 3, (255, 0, 0), -1)
-
-            import math
-            # 绘制短轴（蓝色线条）
-            minor_length = axes_minor / 2
-            minor_angle_rad = math.radians(minor_angle)
-            dx_minor = minor_length * math.cos(minor_angle_rad)
-            dy_minor = minor_length * math.sin(minor_angle_rad)
-            pt1_minor = (int(cx_ell - dx_minor), int(cy_ell - dy_minor))
-            pt2_minor = (int(cx_ell + dx_minor), int(cy_ell + dy_minor))
-            cv2.line(img_cv, pt1_minor, pt2_minor, (0, 0, 255), 2)  # 蓝色 (RGB格式)
-        else:
-            # 如果没有椭圆参数，绘制圆形（红色）
-            cv2.circle(img_cv, (cx, cy), radius, (0, 0, 255), 2)
-            cv2.circle(img_cv, (cx, cy), 2, (0, 0, 255), -1)
-
-        # 转换回 maix image
-        result_img = image.cv2image(img_cv, False, False)
-
-        # 定义颜色对象用于文字
-        try:
-            color_black = image.Color.from_rgb(0, 0, 0)
-        except AttributeError:
-            color_black = image.Color(0, 0, 0)
-
-        # D. 添加文字信息
-        FOCAL_LENGTH_PIX = 1900
-        d = (REAL_RADIUS_CM * FOCAL_LENGTH_PIX) / radius1 / 100.0
-        info_str = f"R:{radius} M:{method} D:{d:.2f}"
-        print(info_str)
-
-        # 计算文字位置，防止超出图片边界
-        r_outer = int(radius * 11.0) if radius else 100
-        text_y = cy - r_outer - 20 if cy > r_outer + 20 else cy + r_outer + 20
-
-        # 调用 draw_string
-        result_img.draw_string(0, 0, info_str, color=color_black, scale=1.0)
-
-    # 5. 保存结果图片
-    base, ext = os.path.splitext(image_path)
-    output_path = f"{base}_result{ext}"
-    try:
-        result_img.save(output_path, quality=100)
-        print(f"[SUCCESS] 结果已保存至: {output_path}")
-    except Exception as e:
-        print(f"[ERROR] 保存图片失败: {e}")
-
-
-if __name__ == "__main__":
-    # ================= 配置区域 =================
-
-    # 1. 设置要测试的图片路径
-    # 建议将图片放在与脚本同级目录，或者使用绝对路径
-    TARGET_IMAGE = "/root/phot/None_314_258_0_0041.bmp"
-
-    TARGET_DIR = "/root/phot"  # 修改为你想要读取的目录路径
-
-    # 支持的图片格式
-    IMAGE_EXTENSIONS = ['.jpg', '.jpeg', '.png', '.bmp']
-
-    # ================= 执行区域 =================
-    if 'TARGET_DIR' in locals():
-        # 读取目录下所有图片文件，过滤掉 _result.jpg 后缀的文件
-        image_files = []
-        if os.path.exists(TARGET_DIR) and os.path.isdir(TARGET_DIR):
-            for filename in os.listdir(TARGET_DIR):
-                # 检查文件扩展名
-                if any(filename.lower().endswith(ext) for ext in IMAGE_EXTENSIONS):
-                    # 过滤掉 _result.jpg 后缀的文件
-                    if not filename.endswith('_result.jpg'):
-                        filepath = os.path.join(TARGET_DIR, filename)
-                        if os.path.isfile(filepath):
-                            image_files.append(filepath)
-
-            # 按文件名排序（可选）
-            image_files.sort()
-
-            print(f"[INFO] 在目录 {TARGET_DIR} 中找到 {len(image_files)} 张图片")
-
-            # 处理每张图片
-            for img_path in image_files:
-                print(f"\n{'=' * 10} 开始处理: {img_path} {'=' * 10}")
-                run_offline_test(img_path)
-        else:
-            print(f"[ERROR] 目录不存在或不是有效目录: {TARGET_DIR}")
-
-    else:
-        run_offline_test(TARGET_IMAGE)

test/test_decect_circle_v4.py

@@ -1,635 +0,0 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-离线测试脚本：直接复用 detect_circle 逻辑进行测试
-运行环境：MaixPy (Sipeed MAIX)
-"""
-import sys
-import os
-# import time
-from maix import image, time
-import cv2
-import numpy as np
-
-# ==================== 全局配置 (与 test_main.py 保持一致) ====================
-REAL_RADIUS_CM = 20  # 靶心实际半径（厘米）
-
-
-# ==================== 复制的核心算法 ====================
-# 注意：这里直接复制了 detect_circle 的逻辑，避免 import main 导致的冲突
-
-
-def detect_circle_v3(frame, laser_point=None):
-    """检测图像中的靶心（优先清晰轮廓，其次黄色区域）- 返回椭圆参数版本
-    增加红色圆圈检测，验证黄色圆圈是否为真正的靶心
-    如果提供 laser_point，会选择最接近激光点的目标
-    
-    Args:
-        frame: 图像帧
-        laser_point: 激光点坐标 (x, y)，用于多目标场景下的目标选择
-    
-    Returns:
-        (result_img, best_center, best_radius, method, best_radius1, ellipse_params)
-    """
-    img_cv = image.image2cv(frame, False, False)
-
-    best_center = best_radius = best_radius1 = method = None
-    ellipse_params = None
-
-    # HSV 黄色掩码检测（模糊靶心）
-    hsv = cv2.cvtColor(img_cv, cv2.COLOR_RGB2HSV)
-    h, s, v = cv2.split(hsv)
-
-    # 调整饱和度策略：稍微增强，不要过度
-    s = np.clip(s * 1.1, 0, 255).astype(np.uint8)
-
-    hsv = cv2.merge((h, s, v))
-
-    # 放宽 HSV 阈值范围（针对模糊图像的关键调整）
-    lower_yellow = np.array([7, 80, 0])  # 饱和度下限降低，捕捉淡黄色
-    upper_yellow = np.array([32, 255, 255])  # 亮度上限拉满
-
-    mask_yellow = cv2.inRange(hsv, lower_yellow, upper_yellow)
-
-    # 调整形态学操作
-    kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
-    mask_yellow = cv2.morphologyEx(mask_yellow, cv2.MORPH_CLOSE, kernel)
-
-    contours_yellow, _ = cv2.findContours(mask_yellow, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-
-    # 存储所有有效的黄色-红色组合
-    valid_targets = []
-
-    if contours_yellow:
-        for cnt_yellow in contours_yellow:
-            area = cv2.contourArea(cnt_yellow)
-            perimeter = cv2.arcLength(cnt_yellow, True)
-
-            # 计算圆度
-            if perimeter > 0:
-                circularity = (4 * np.pi * area) / (perimeter * perimeter)
-            else:
-                circularity = 0
-
-            if area > 50 and circularity > 0.7:
-                print(f"[target] -> 面积:{area}, 圆度:{circularity:.2f}")
-                # 尝试拟合椭圆
-                yellow_center = None
-                yellow_radius = None
-                yellow_ellipse = None
-
-                if len(cnt_yellow) >= 5:
-                    (x, y), (width, height), angle = cv2.fitEllipse(cnt_yellow)
-                    yellow_ellipse = ((x, y), (width, height), angle)
-                    axes_minor = min(width, height)
-                    radius = axes_minor / 2
-                    yellow_center = (int(x), int(y))
-                    yellow_radius = int(radius)
-                else:
-                    (x, y), radius = cv2.minEnclosingCircle(cnt_yellow)
-                    yellow_center = (int(x), int(y))
-                    yellow_radius = int(radius)
-                    yellow_ellipse = None
-
-                # 如果检测到黄色圆圈，再检测红色圆圈进行验证
-                if yellow_center and yellow_radius:
-                    # HSV 红色掩码检测（红色在HSV中跨越0度，需要两个范围）
-                    # 红色范围1: 0-12度（接近0度的红色）
-                    # 放宽S/V阈值：S>=30, V>=20 以捕获淡红/暗红
-                    lower_red1 = np.array([0, 30, 20])
-                    upper_red1 = np.array([12, 255, 255])
-                    mask_red1 = cv2.inRange(hsv, lower_red1, upper_red1)
-
-                    # 红色范围2: 168-180度（接近180度的红色）
-                    lower_red2 = np.array([168, 30, 20])
-                    upper_red2 = np.array([180, 255, 255])
-                    mask_red2 = cv2.inRange(hsv, lower_red2, upper_red2)
-
-                    # 合并两个红色掩码
-                    mask_red = cv2.bitwise_or(mask_red1, mask_red2)
-
-                    # 形态学操作：先CLOSE填充空洞，再DILATE加厚环状区域
-                    kernel_red = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
-                    mask_red = cv2.morphologyEx(mask_red, cv2.MORPH_CLOSE, kernel_red)
-                    mask_red = cv2.dilate(mask_red, kernel_red, iterations=1)
-
-                    contours_red, _ = cv2.findContours(mask_red, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-
-                    red_pixel_count = np.sum(mask_red > 0)
-                    print(f"Debug -> 红色掩码: {red_pixel_count} 像素, {len(contours_red)} 个轮廓")
-
-                    found_valid_red = False
-
-                    if contours_red:
-                        for cnt_red in contours_red:
-                            area_red = cv2.contourArea(cnt_red)
-                            perimeter_red = cv2.arcLength(cnt_red, True)
-
-                            if perimeter_red > 0:
-                                circularity_red = (4 * np.pi * area_red) / (perimeter_red * perimeter_red)
-                            else:
-                                circularity_red = 0
-
-                            # 环状轮廓圆度可能偏低，放宽到0.2
-                            print(f"Debug -> 红轮廓: 面积={area_red:.1f}, 圆度={circularity_red:.2f}" +
-                                  f" (面积>15={area_red > 15}, 圆度>0.2={circularity_red > 0.2})")
-                            if area_red > 15 and circularity_red > 0.2:
-                                if len(cnt_red) >= 5:
-                                    (x_red, y_red), (w_red, h_red), angle_red = cv2.fitEllipse(cnt_red)
-                                    radius_red = min(w_red, h_red) / 2
-                                    red_center = (int(x_red), int(y_red))
-                                    red_radius = int(radius_red)
-                                else:
-                                    (x_red, y_red), radius_red = cv2.minEnclosingCircle(cnt_red)
-                                    red_center = (int(x_red), int(y_red))
-                                    red_radius = int(radius_red)
-
-                                if red_center:
-                                    dx = yellow_center[0] - red_center[0]
-                                    dy = yellow_center[1] - red_center[1]
-                                    distance = np.sqrt(dx * dx + dy * dy)
-
-                                    max_distance = yellow_radius * 2.0
-                                    min_r = min(red_radius, yellow_radius)
-                                    max_r = max(red_radius, yellow_radius)
-                                    size_ratio = min_r / max_r if max_r > 0 else 0
-                                    print(f"Debug -> 圆心距={distance:.1f}(阈值={max_distance:.1f}), "
-                                          f"大小比={size_ratio:.2f}(阈值=0.5), "
-                                          f"距离OK={distance < max_distance}, 大小OK={size_ratio > 0.5}")
-
-                                    # 允许红圈在黄圈外侧或内侧，只要大小相近（较小/较大 >= 0.5）
-                                    if distance < max_distance and size_ratio > 0.5:
-                                        found_valid_red = True
-                                        print(
-                                            f"[target] -> 找到匹配的红圈: 黄心({yellow_center}), 红心({red_center}), 距离:{distance:.1f}, 黄半径:{yellow_radius}, 红半径:{red_radius}")
-
-                                        valid_targets.append({
-                                            'center': yellow_center,
-                                            'radius': yellow_radius,
-                                            'ellipse': yellow_ellipse,
-                                            'area': area
-                                        })
-                                        break
-
-                    if not found_valid_red:
-                        # 如果黄圈非常可靠（大且圆），在没有红圈验证时仍接受
-                        if area > 30 and circularity > 0.85:
-                            print(f"[target] -> 黄圈高置信度(面积:{area:.0f}, 圆度:{circularity:.2f})，跳过红圈验证直接接受")
-                            valid_targets.append({
-                                'center': yellow_center,
-                                'radius': yellow_radius,
-                                'ellipse': yellow_ellipse,
-                                'area': area
-                            })
-                        else:
-                            print("Debug -> 未找到匹配的红色圆圈，可能是误识别")
-
-    # 从所有有效目标中选择最佳目标
-    if valid_targets:
-        if laser_point:
-            # 如果有激光点，选择最接近激光点的目标
-            best_target = None
-            min_distance = float('inf')
-            for target in valid_targets:
-                dx = target['center'][0] - laser_point[0]
-                dy = target['center'][1] - laser_point[1]
-                distance = np.sqrt(dx * dx + dy * dy)
-                if distance < min_distance:
-                    min_distance = distance
-                    best_target = target
-            if best_target:
-                best_center = best_target['center']
-                best_radius = best_target['radius']
-                ellipse_params = best_target['ellipse']
-                method = "v3_ellipse_red_validated_laser_selected"
-                best_radius1 = best_radius * 5
-        else:
-            # 如果没有激光点，选择面积最大的目标
-            best_target = max(valid_targets, key=lambda t: t['area'])
-            best_center = best_target['center']
-            best_radius = best_target['radius']
-            ellipse_params = best_target['ellipse']
-            method = "v3_ellipse_red_validated"
-            best_radius1 = best_radius * 5
-
-    result_img = image.cv2image(img_cv, False, False)
-    return result_img, best_center, best_radius, method, best_radius1, ellipse_params
-
-
-def detect_circle(frame):
-    """检测图像中的靶心（优先清晰轮廓，其次黄色区域）"""
-    img_cv = image.image2cv(frame, False, False)
-    # gray = cv2.cvtColor(img_cv, cv2.COLOR_RGB2GRAY)
-    # blurred = cv2.GaussianBlur(gray, (5, 5), 0)
-    # edged = cv2.Canny(blurred, 50, 150)
-    # kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
-    # ceroded = cv2.erode(cv2.dilate(edged, kernel), kernel)
-
-    # contours, _ = cv2.findContours(ceroded, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
-    # best_center = best_radius = best_radius1 = method = None
-
-    # hsv = cv2.cvtColor(img_cv, cv2.COLOR_RGB2HSV)
-    # h, s, v = cv2.split(hsv)
-    # s = np.clip(s * 2, 0, 255).astype(np.uint8)
-    # hsv = cv2.merge((h, s, v))
-    # lower_yellow = np.array([7, 80, 0])
-    # upper_yellow = np.array([32, 255, 182])
-    # mask = cv2.inRange(hsv, lower_yellow, upper_yellow)
-    # kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
-    # mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, kernel)
-    # mask = cv2.morphologyEx(mask, cv2.MORPH_DILATE, kernel)
-    # contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-    # if contours:
-    #     largest = max(contours, key=cv2.contourArea)
-    #     if cv2.contourArea(largest) > 50:
-    #         (x, y), radius = cv2.minEnclosingCircle(largest)
-    #         best_center = (int(x), int(y))
-    #         best_radius = int(radius)
-    #         best_radius1 = radius * 5
-    #         method = "v2"
-
-    # auto
-    # R:31 M:v2 D:2.410110127692767
-    # hsv = cv2.cvtColor(img_cv, cv2.COLOR_RGB2HSV)
-    # h, s, v = cv2.split(hsv)
-
-    # # 1. 增强饱和度（模糊照片需要更强的增强）
-    # s = np.clip(s * 2.5, 0, 255).astype(np.uint8)  # 从2.0改为2.5
-
-    # # 2. 增强亮度（模糊照片可能偏暗）
-    # v = np.clip(v * 1.2, 0, 255).astype(np.uint8)  # 新增：提升亮度
-
-    # hsv = cv2.merge((h, s, v))
-
-    # # 3. 放宽HSV颜色范围（特别是模糊照片）
-    # # 降低饱和度下限，提高亮度上限
-    # lower_yellow = np.array([5, 50, 30])   # H:5-35, S:50-255, V:30-255
-    # upper_yellow = np.array([35, 255, 255])
-
-    # mask = cv2.inRange(hsv, lower_yellow, upper_yellow)
-
-    # # 4. 增强形态学操作（连接被分割的区域）
-    # kernel_small = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
-    # kernel_large = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (9, 9))  # 更大的核
-
-    # # 先开运算去除噪声
-    # mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, kernel_small)
-    # # 多次膨胀连接区域（模糊照片需要更多膨胀）
-    # mask = cv2.dilate(mask, kernel_large, iterations=2)  # 增加迭代次数
-    # mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, kernel_large)  # 闭运算填充空洞
-
-    # contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-    # if contours:
-    #     largest = max(contours, key=cv2.contourArea)
-    #     area = cv2.contourArea(largest)
-    #     if area > 50:
-    #         # 5. 使用面积计算等效半径（更准确）
-    #         equivalent_radius = np.sqrt(area / np.pi)
-
-    #         # 6. 同时使用minEnclosingCircle作为备选（取较大值）
-    #         (x, y), enclosing_radius = cv2.minEnclosingCircle(largest)
-
-    #         # 取两者中的较大值，确保不遗漏
-    #         radius = max(equivalent_radius, enclosing_radius)
-
-    #         best_center = (int(x), int(y))
-    #         best_radius = int(radius)
-    #         best_radius1 = radius * 5
-    #         method = "v2"
-
-    # codegee
-    # R:24 M:v2 D:3.061493895819174
-    # R:22 M:v2 D:3.3644971681267077  np.clip(s * 1.1, 0, 255)
-    hsv = cv2.cvtColor(img_cv, cv2.COLOR_RGB2HSV)
-    h, s, v = cv2.split(hsv)
-
-    # 2. 调整饱和度策略：
-    # 不要暴力翻倍，可以尝试稍微增强，或者使用 CLAHE 增强亮度/对比度
-    # 这里我们稍微增加一点饱和度，并确保不溢出
-    s = np.clip(s * 1.1, 0, 255).astype(np.uint8)
-    # 对亮度通道 v 也可以做一点 CLAHE 处理来增强对比度（可选）
-    # clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8,8))
-    # v = clahe.apply(v)
-
-    hsv = cv2.merge((h, s, v))
-
-    # 3. 放宽 HSV 阈值范围（针对模糊图像的关键调整）
-    # 降低 S 的下限 (80 -> 35)，提高 V 的上限 (182 -> 255)
-    lower_yellow = np.array([7, 80, 0])  # 饱和度下限降低，捕捉淡黄色
-    upper_yellow = np.array([32, 255, 255])  # 亮度上限拉满
-
-    mask = cv2.inRange(hsv, lower_yellow, upper_yellow)
-
-    # 4. 调整形态学操作
-    # 去掉 MORPH_OPEN，因为它会减小面积。
-    # 使用 MORPH_CLOSE (先膨胀后腐蚀) 来填充内部小黑洞，连接近邻区域
-    kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
-    mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, kernel)
-    # 再进行一次膨胀，确保边缘被包含进来
-    # mask = cv2.dilate(mask, kernel, iterations=1) 
-
-    contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-
-    if contours:
-        largest = max(contours, key=cv2.contourArea)
-
-        # 这里可以适当降低面积阈值，或者保持不变
-        if cv2.contourArea(largest) > 50:
-            # (x, y), radius = cv2.minEnclosingCircle(largest)
-            # best_center = (int(x), int(y))
-            # best_radius = int(radius)
-
-            # --- 核心修改开始 ---
-            # 1. 尝试拟合椭圆 (需要轮廓点至少为5个)
-            if len(largest) >= 5:
-                # 返回值: ((中心x, 中心y), (长轴, 短轴), 旋转角度)
-                (x, y), (axes_major, axes_minor), angle = cv2.fitEllipse(largest)
-
-                # 2. 计算半径
-                # 选项A：取长短轴的平均值 (比较稳健)
-                # radius = (axes_major + axes_minor) / 4
-
-                # 选项B：直接取短轴的一半 (抗模糊最强，推荐)
-                radius = axes_minor / 2
-
-                best_center = (int(x), int(y))
-                best_radius = int(radius)
-                method = "v2_ellipse"
-            else:
-                # 如果点太少无法拟合椭圆，降级回 minEnclosingCircle
-                (x, y), radius = cv2.minEnclosingCircle(largest)
-                best_center = (int(x), int(y))
-                best_radius = int(radius)
-                method = "v2"
-            # --- 核心修改结束 ---
-
-            # 你的后续逻辑
-            best_radius1 = radius * 5
-
-    # operas 4.5
-    # R:25 M:v2 D:2.9554872521538527
-    # hsv = cv2.cvtColor(img_cv, cv2.COLOR_RGB2HSV)
-    # h, s, v = cv2.split(hsv)
-
-    # # 1. 适度增强饱和度（不要过度，否则噪声也会增强）
-    # s = np.clip(s * 1.5, 0, 255).astype(np.uint8)
-    # hsv = cv2.merge((h, s, v))
-
-    # # 2. 放宽 HSV 阈值范围（关键改动）
-    # # - 饱和度下限从 80 降到 40（捕捉淡黄色）
-    # # - 亮度上限从 182 提高到 255（允许更亮的黄色）
-    # lower_yellow = np.array([7, 40, 30])
-    # upper_yellow = np.array([35, 255, 255])
-
-    # mask = cv2.inRange(hsv, lower_yellow, upper_yellow)
-
-    # # 3. 调整形态学操作：用 CLOSE 替代 OPEN
-    # # CLOSE（先膨胀后腐蚀）：填充内部空洞，连接相邻区域
-    # # OPEN（先腐蚀后膨胀）：会缩小区域，不适合模糊图像
-    # kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (7, 7))  # 稍大的核
-    # mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, kernel)
-    # mask = cv2.dilate(mask, kernel, iterations=1)  # 额外膨胀，确保边缘被包含
-
-    # contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-    # if contours:
-    #     largest = max(contours, key=cv2.contourArea)
-    #     if cv2.contourArea(largest) > 50:
-    #         (x, y), radius = cv2.minEnclosingCircle(largest)
-    #         best_center = (int(x), int(y))
-    #         best_radius = int(radius)
-    #         best_radius1 = radius * 5
-    #         method = "v2"
-
-    # # --- 新增：将 Mask 叠加到原图上用于调试 ---
-    # # 创建一个彩色掩码（红色通道为255，其他为0）
-    # mask_overlay = np.zeros_like(img_cv)
-    # mask_overlay[:, :, 2] = mask  # 将掩码放在红色通道 (BGR中的R)
-    #
-    # cv2.addWeighted(img_cv, 0.6, mask_overlay, 0.4, 0, img_cv)
-
-    result_img = image.cv2image(img_cv, False, False)
-    return result_img, best_center, best_radius, method, best_radius1
-
-
-def detect_circle_v2(frame):
-    """检测图像中的靶心（优先清晰轮廓，其次黄色区域）- 返回椭圆参数版本"""
-    global REAL_RADIUS_CM
-    img_cv = image.image2cv(frame, False, False)
-
-    best_center = best_radius = best_radius1 = method = None
-    ellipse_params = None  # 存储椭圆参数 ((x, y), (axes_major, axes_minor), angle)
-
-    # HSV 黄色掩码检测（模糊靶心）
-    hsv = cv2.cvtColor(img_cv, cv2.COLOR_RGB2HSV)
-    h, s, v = cv2.split(hsv)
-
-    # 调整饱和度策略：稍微增强，不要过度
-    s = np.clip(s * 1.1, 0, 255).astype(np.uint8)
-
-    hsv = cv2.merge((h, s, v))
-
-    # 放宽 HSV 阈值范围（针对模糊图像的关键调整）
-    lower_yellow = np.array([7, 80, 0])  # 饱和度下限降低，捕捉淡黄色
-    upper_yellow = np.array([32, 255, 255])  # 亮度上限拉满
-
-    mask = cv2.inRange(hsv, lower_yellow, upper_yellow)
-
-    # 调整形态学操作
-    # 使用 MORPH_CLOSE (先膨胀后腐蚀) 来填充内部小黑洞，连接近邻区域
-    kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
-    mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, kernel)
-
-    contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-
-    if contours:
-        largest = max(contours, key=cv2.contourArea)
-
-        if cv2.contourArea(largest) > 50:
-            # 尝试拟合椭圆 (需要轮廓点至少为5个)
-            if len(largest) >= 5:
-                # 返回值: ((中心x, 中心y), (width, height), 旋转角度)
-                # 注意：width 和 height 是外接矩形的尺寸，不是长轴和短轴
-                (x, y), (width, height), angle = cv2.fitEllipse(largest)
-
-                # 保存椭圆参数（保持原始顺序，用于绘制）
-                ellipse_params = ((x, y), (width, height), angle)
-
-                # 计算半径：使用较小的尺寸作为短轴
-                axes_minor = min(width, height)
-                radius = axes_minor / 2
-
-                best_center = (int(x), int(y))
-                best_radius = int(radius)
-                method = "v2_ellipse"
-            else:
-                # 如果点太少无法拟合椭圆，降级回 minEnclosingCircle
-                (x, y), radius = cv2.minEnclosingCircle(largest)
-                best_center = (int(x), int(y))
-                best_radius = int(radius)
-                method = "v2"
-                ellipse_params = None  # 圆形，没有椭圆参数
-
-            best_radius1 = radius * 5
-
-    result_img = image.cv2image(img_cv, False, False)
-    return result_img, best_center, best_radius, method, best_radius1, ellipse_params
-
-
-# ==================== 测试逻辑 ====================
-
-def run_offline_test(image_path):
-    """读取图片，检测圆，绘制结果，保存图片"""
-
-    # 1. 检查文件是否存在
-    if not os.path.exists(image_path):
-        print(f"[ERROR] 找不到图片文件: {image_path}")
-        return
-
-    # 2. 使用 maix.image 读取图片 (适配 MaixPy v4)
-    try:
-        # 使用 image.load 读取文件，返回 Image 对象
-        img = image.load(image_path)
-        print(f"[INFO] 成功读取图片: {image_path} (尺寸: {img.width()}x{img.height()})")
-    except Exception as e:
-        print(f"[ERROR] 读取图片失败: {e}")
-        print("提示：请确认 MaixPy 版本是否为 v4，且图片路径正确。")
-        return
-
-    # 3. 调用 detect_circle_v2 函数
-    print("[INFO] 正在调用 detect_circle_v2 进行检测...")
-    start_time = time.ticks_ms()
-
-    result_img, center, radius, method, radius1, ellipse_params = detect_circle_v3(img)
-
-    cost_time = time.ticks_ms() - start_time
-    print(f"[INFO] 检测完成，耗时: {cost_time}ms")
-    print(f"       结果 -> 圆心: {center}, 半径: {radius}, 方法: {method}")
-    if ellipse_params:
-        (ell_center, (width, height), angle) = ellipse_params
-        print(
-            f"       椭圆 -> 中心: ({ell_center[0]:.1f}, {ell_center[1]:.1f}), 长轴: {max(width, height):.1f}, 短轴: {min(width, height):.1f}, 角度: {angle:.1f}°")
-
-    # 4. 绘制辅助线（可选，用于调试）
-    if center and radius:
-        # 为了绘制椭圆，需要转换回 cv2 图像
-        img_cv = image.image2cv(result_img, False, False)
-
-        cx, cy = center
-
-        # 如果有椭圆参数，绘制椭圆
-        if ellipse_params:
-            (ell_center, (width, height), angle) = ellipse_params
-            cx_ell, cy_ell = int(ell_center[0]), int(ell_center[1])
-
-            # 确定长轴和短轴
-            if width >= height:
-                # width 是长轴，height 是短轴
-                axes_major = width
-                axes_minor = height
-                major_angle = angle  # 长轴角度就是 angle
-                minor_angle = angle + 90  # 短轴角度 = 长轴角度 + 90度
-            else:
-                # height 是长轴，width 是短轴
-                axes_major = height
-                axes_minor = width
-                major_angle = angle + 90  # 长轴角度 = width角度 + 90度
-                minor_angle = angle  # 短轴角度就是 angle
-
-            # 使用 OpenCV 绘制椭圆（绿色，线宽2）
-            cv2.ellipse(img_cv,
-                        (cx_ell, cy_ell),  # 中心点
-                        (int(width / 2), int(height / 2)),  # 半宽、半高
-                        angle,  # 旋转角度（OpenCV需要原始angle）
-                        0, 360,  # 起始和结束角度
-                        (0, 255, 0),  # 绿色 (RGB格式)
-                        2)  # 线宽
-
-            # 绘制椭圆中心点（红色）
-            cv2.circle(img_cv, (cx_ell, cy_ell), 3, (255, 0, 0), -1)
-
-            import math
-            # 绘制短轴（蓝色线条）
-            minor_length = axes_minor / 2
-            minor_angle_rad = math.radians(minor_angle)
-            dx_minor = minor_length * math.cos(minor_angle_rad)
-            dy_minor = minor_length * math.sin(minor_angle_rad)
-            pt1_minor = (int(cx_ell - dx_minor), int(cy_ell - dy_minor))
-            pt2_minor = (int(cx_ell + dx_minor), int(cy_ell + dy_minor))
-            cv2.line(img_cv, pt1_minor, pt2_minor, (0, 0, 255), 2)  # 蓝色 (RGB格式)
-        else:
-            # 如果没有椭圆参数，绘制圆形（红色）
-            cv2.circle(img_cv, (cx, cy), radius, (0, 0, 255), 2)
-            cv2.circle(img_cv, (cx, cy), 2, (0, 0, 255), -1)
-
-        # 转换回 maix image
-        result_img = image.cv2image(img_cv, False, False)
-
-        # 定义颜色对象用于文字
-        try:
-            color_black = image.Color.from_rgb(0, 0, 0)
-        except AttributeError:
-            color_black = image.Color(0, 0, 0)
-
-        # D. 添加文字信息
-        FOCAL_LENGTH_PIX = 1900
-        d = (REAL_RADIUS_CM * FOCAL_LENGTH_PIX) / radius1 / 100.0
-        info_str = f"R:{radius} M:{method} D:{d:.2f}"
-        print(info_str)
-
-        # 计算文字位置，防止超出图片边界
-        r_outer = int(radius * 11.0) if radius else 100
-        text_y = cy - r_outer - 20 if cy > r_outer + 20 else cy + r_outer + 20
-
-        # 调用 draw_string
-        result_img.draw_string(0, 0, info_str, color=color_black, scale=1.0)
-
-    # 5. 保存结果图片
-    output_path = image_path.replace(".bmp", "_result.bmp")
-    output_path = image_path.replace(".jpg", "_result.jpg")
-    try:
-        result_img.save(output_path, quality=100)
-        print(f"[SUCCESS] 结果已保存至: {output_path}")
-    except Exception as e:
-        print(f"[ERROR] 保存图片失败: {e}")
-
-
-if __name__ == "__main__":
-    # ================= 配置区域 =================
-
-    # 1. 设置要测试的图片路径
-    # 建议将图片放在与脚本同级目录，或者使用绝对路径
-    TARGET_IMAGE = "/root/phot/None_314_258_0_0041.bmp"
-
-    TARGET_DIR = "/root/phot"  # 修改为你想要读取的目录路径
-
-    # 支持的图片格式
-    IMAGE_EXTENSIONS = ['.jpg', '.jpeg', '.png', '.bmp']
-
-    # ================= 执行区域 =================
-    if 'TARGET_DIR' in locals():
-        # 读取目录下所有图片文件，过滤掉 _result.jpg 后缀的文件
-        image_files = []
-        if os.path.exists(TARGET_DIR) and os.path.isdir(TARGET_DIR):
-            for filename in os.listdir(TARGET_DIR):
-                # 检查文件扩展名
-                if any(filename.lower().endswith(ext) for ext in IMAGE_EXTENSIONS):
-                    # 过滤掉 _result.jpg 后缀的文件
-                    if filename.endswith('no_target.jpg'):
-                        filepath = os.path.join(TARGET_DIR, filename)
-                        if os.path.isfile(filepath):
-                            image_files.append(filepath)
-
-            # 按文件名排序（可选）
-            image_files.sort()
-
-            print(f"[INFO] 在目录 {TARGET_DIR} 中找到 {len(image_files)} 张图片")
-
-            # 处理每张图片
-            for img_path in image_files:
-                print(f"\n{'=' * 10} 开始处理: {img_path} {'=' * 10}")
-                run_offline_test(img_path)
-        else:
-            print(f"[ERROR] 目录不存在或不是有效目录: {TARGET_DIR}")
-
-    else:
-        run_offline_test(TARGET_IMAGE)

test/test_laser_center_point.py

@@ -0,0 +1,541 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+激光中心点检测单元测试（单文件，无项目依赖）
+直接使用 maix 标准库，实现红色激光点坐标检测
+
+运行方式：
+  python3 test/test_laser_center_point.py
+
+Ctrl+C 退出，按 s 保存截图
+"""
+
+from maix import camera, display, image, time, app, uart, pinmap
+import os
+import struct
+import select
+
+_USE_CV = False
+try:
+    import cv2
+    import numpy as np
+    _USE_CV = True
+except ImportError:
+    pass
+
+WIDTH = 640
+HEIGHT = 480
+THRESHOLD = 140
+SEARCH_RADIUS = 50
+
+
+def read_key_ev():
+    """非阻塞读取 /dev/input/event0 按键（返回 key_code 或 -1）"""
+    try:
+        r, _, _ = select.select([_key_fd], [], [], 0)
+        if r:
+            event = _key_fd.read(16)
+            if len(event) == 16:
+                _, _, etype, code, value = struct.unpack("IIHHI", event)
+                if etype == 1 and value == 1:
+                    return code
+    except Exception:
+        pass
+    return -1
+
+
+def find_ellipse(img_cv, cx, cy, roi_r, th):
+    x1 = max(0, cx - roi_r)
+    x2 = min(WIDTH, cx + roi_r)
+    y1 = max(0, cy - roi_r)
+    y2 = min(HEIGHT, cy + roi_r)
+    roi = img_cv[y1:y2, x1:x2]
+    if roi.size == 0:
+        return None
+    r = roi[:, :, 0].astype(np.int32)
+    g = roi[:, :, 1].astype(np.int32)
+    b = roi[:, :, 2].astype(np.int32)
+    mask = (r > th) & (r > g * 1.5) & (r > b * 1.5)
+    oe = (r > 200) & (g > 200) & (b > 200) & (r >= g) & (r >= b) & ((r - g) > 10) & ((r - b) > 10)
+    combined = (mask | oe).astype(np.uint8) * 255
+    contours, _ = cv2.findContours(combined, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+    if not contours:
+        return None
+    largest = max(contours, key=cv2.contourArea)
+    if cv2.contourArea(largest) < 5:
+        return None
+    cnt = largest.copy()
+    for pt in cnt:
+        pt[0][0] += x1
+        pt[0][1] += y1
+    if len(cnt) >= 5:
+        (ex, ey), (ew, eh), ang = cv2.fitEllipse(cnt)
+        mask_ellipse = np.zeros((HEIGHT, WIDTH), dtype=np.uint8)
+        cv2.ellipse(mask_ellipse, (int(ex), int(ey)), (int(ew / 2), int(eh / 2)), ang, 0, 360, 255, -1)
+        brightness = img_cv[:, :, 0].astype(np.int32) + img_cv[:, :, 1].astype(np.int32) + img_cv[:, :, 2].astype(np.int32)
+        masked = np.where(mask_ellipse > 0, brightness, 0)
+        vals = masked[masked > 0]
+        if len(vals) > 0:
+            bth = np.percentile(vals, 90)
+            bmask = (masked >= bth).astype(np.uint8) * 255
+            bcontours, _ = cv2.findContours(bmask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+            if bcontours:
+                blargest = max(bcontours, key=cv2.contourArea)
+                if cv2.contourArea(blargest) >= 3 and len(blargest) >= 5:
+                    (ix, iy), _, _ = cv2.fitEllipse(blargest)
+                    return (float(ix), float(iy))
+                M = cv2.moments(blargest)
+                if M["m00"] > 0:
+                    return (float(M["m10"] / M["m00"]), float(M["m01"] / M["m00"]))
+        return (float(ex), float(ey))
+    M = cv2.moments(cnt)
+    if M["m00"] > 0:
+        return (float(M["m10"] / M["m00"]), float(M["m01"] / M["m00"]))
+    return None
+
+
+def find_brightest(img_cv, cx, cy, roi_r, th):
+    x1 = max(0, cx - roi_r)
+    x2 = min(WIDTH, cx + roi_r)
+    y1 = max(0, cy - roi_r)
+    y2 = min(HEIGHT, cy + roi_r)
+    best_score = 0
+    best_pos = None
+    for y in range(y1, y2):
+        for x in range(x1, x2):
+            r, g, b = int(img_cv[y, x, 0]), int(img_cv[y, x, 1]), int(img_cv[y, x, 2])
+            is_red = (r > th and r > g * 1.5 and r > b * 1.5)
+            is_oe = (r > 200 and g > 200 and b > 200 and r >= g and r >= b and (r - g) > 10 and (r - b) > 10)
+            if is_red or is_oe:
+                score = r + g + b
+                dx, dy = x - cx, y - cy
+                dist = (dx * dx + dy * dy) ** 0.5
+                score *= max(0.5, 1.0 - (dist / roi_r) * 0.5)
+                if score > best_score:
+                    best_score = score
+                    best_pos = (float(x), float(y))
+    return best_pos
+
+
+# 打开键盘输入设备
+_key_fd = None
+try:
+    _key_fd = open("/dev/input/event0", "rb")
+except Exception:
+    try:
+        _key_fd = open("/dev/input/event1", "rb")
+    except Exception:
+        _key_fd = None
+
+print("=" * 50)
+print("激光中心点检测单元测试")
+print("=" * 50)
+print()
+
+cam = camera.Camera(WIDTH, HEIGHT)
+disp = display.Display()
+print("[OK] 摄像头和显示初始化完成")
+
+# 初始化激光串口
+_laser_on = False
+_laser_uart = None
+try:
+    pinmap.set_pin_function("A18", "UART1_RX")
+    pinmap.set_pin_function("A19", "UART1_TX")
+    _laser_uart = uart.UART("/dev/ttyS1", 9600)
+    _laser_uart.read(-1)
+    print("[OK] 激光串口初始化完成")
+except Exception as e:
+    print(f"[WARN] 激光串口初始化失败: {e}")
+
+LASER_ON = bytes([0xAA, 0x00, 0x01, 0xBE, 0x00, 0x01, 0x00, 0x01, 0xC1])
+LASER_OFF = bytes([0xAA, 0x00, 0x01, 0xBE, 0x00, 0x01, 0x00, 0x00, 0xC0])
+
+# 默认开启激光
+if _laser_uart:
+    try:
+        _laser_uart.write(LASER_ON)
+        time.sleep_ms(50)
+        _laser_uart.read(-1)
+        _laser_on = True
+        print("[OK] 激光已开启")
+    except Exception as e:
+        print(f"[WARN] 开启激光失败: {e}")
+print()
+
+pos_ellipse = None
+pos_bright = None
+frame_count = 0
+use_ellipse = True
+
+while not app.need_exit():
+    frame = cam.read()
+    if frame is None:
+        time.sleep_ms(10)
+        continue
+
+    frame_count += 1
+
+    if _USE_CV:
+        img_cv = image.image2cv(frame, False, False)
+        cx, cy = WIDTH // 2, HEIGHT // 2
+
+        t0 = time.ticks_ms()
+        pos_ellipse = find_ellipse(img_cv, cx, cy, SEARCH_RADIUS, THRESHOLD)
+        t1 = time.ticks_ms()
+        pos_bright = find_brightest(img_cv, cx, cy, SEARCH_RADIUS, THRESHOLD)
+        t2 = time.ticks_ms()
+
+        dt_e = abs(time.ticks_diff(t0, t1))
+        dt_b = abs(time.ticks_diff(t1, t2))
+
+        if frame_count % 5 == 0:
+            e_str = f"({pos_ellipse[0]:.1f},{pos_ellipse[1]:.1f})" if pos_ellipse else "None"
+            b_str = f"({pos_bright[0]:.1f},{pos_bright[1]:.1f})" if pos_bright else "None"
+            print(f"[LASER] ellipse={e_str} ({dt_e}ms)  brightest={b_str} ({dt_b}ms)  "
+                  f"th={THRESHOLD}  radius={SEARCH_RADIUS}")
+
+        # 叠加显示
+        pos = pos_ellipse if use_ellipse else pos_bright
+        h, w = img_cv.shape[:2]
+        cv2.circle(img_cv, (cx, cy), SEARCH_RADIUS, (0, 255, 0), 1)
+        cv2.circle(img_cv, (cx, cy), 2, (0, 255, 0), -1)
+        if pos:
+            x, y = int(pos[0]), int(pos[1])
+            cv2.circle(img_cv, (x, y), 6, (0, 0, 255), 2)
+            cv2.line(img_cv, (x - 14, y), (x + 14, y), (0, 0, 255), 1)
+            cv2.line(img_cv, (x, y - 14), (x, y + 14), (0, 0, 255), 1)
+            cv2.putText(img_cv, f"({x},{y})", (x + 10, y - 10),
+                        cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 1, cv2.LINE_AA)
+        info = [
+            f"pos={pos if pos else 'None'}",
+            f"method={'ellipse' if use_ellipse else 'brightest'}  th={THRESHOLD}",
+            f"laser={'ON' if _laser_on else 'OFF'}",
+        ]
+        for i, line in enumerate(info):
+            cv2.putText(img_cv, line, (8, 20 + i * 22),
+                        cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 1, cv2.LINE_AA)
+
+        display_frame = image.cv2image(img_cv, False, False)
+    else:
+        display_frame = frame
+
+    disp.show(display_frame)
+
+    # 按键处理（非阻塞）
+    key = read_key_ev()
+    if key > 0:
+        c = chr(key & 0xFF) if key < 256 else ""
+        if key == 113 or key == 81 or key == 0x1b:  # q/Q/ESC
+            break
+        if c == "e" or key == 18:  # e
+            use_ellipse = not use_ellipse
+            print(f"[KEY] Method: {'ellipse' if use_ellipse else 'brightest'}")
+        if c == "l" or key == 12:  # l
+            _laser_on = not _laser_on
+            if _laser_uart:
+                try:
+                    _laser_uart.write(LASER_ON if _laser_on else LASER_OFF)
+                    time.sleep_ms(30)
+                    _laser_uart.read(-1)
+                    print(f"[KEY] Laser: {'ON' if _laser_on else 'OFF'}")
+                except Exception as e:
+                    print(f"[KEY] Laser error: {e}")
+            else:
+                print("[KEY] Laser UART not available")
+    time.sleep_ms(30)
+
+# 关闭激光
+if _laser_on and _laser_uart:
+    try:
+        _laser_uart.write(LASER_OFF)
+        _laser_uart.read(-1)
+        print("[EXIT] 激光已关闭")
+    except Exception:
+        pass
+print("[EXIT] 测试结束")
+if _key_fd:
+    _key_fd.close()
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+激光中心点检测单元测试（单文件，无项目依赖）
+直接使用 maix 标准库，实现红色激光点坐标检测
+
+运行方式：
+  python3 test/test_laser_center_point.py
+
+Ctrl+C 退出，按 s 保存截图
+"""
+
+from maix import camera, display, image, time, app, uart, pinmap
+import os
+import struct
+import select
+
+_USE_CV = False
+try:
+    import cv2
+    import numpy as np
+    _USE_CV = True
+except ImportError:
+    pass
+
+WIDTH = 640
+HEIGHT = 480
+THRESHOLD = 120
+RED_RATIO = 1.3
+SEARCH_RADIUS = 60
+
+
+def read_key_ev():
+    """非阻塞读取 /dev/input/event0 按键（返回 key_code 或 -1）"""
+    try:
+        r, _, _ = select.select([_key_fd], [], [], 0)
+        if r:
+            event = _key_fd.read(16)
+            if len(event) == 16:
+                _, _, etype, code, value = struct.unpack("IIHHI", event)
+                if etype == 1 and value == 1:
+                    return code
+    except Exception:
+        pass
+    return -1
+
+
+def find_ellipse(img_cv, cx, cy, roi_r, th, ratio):
+    x1 = max(0, cx - roi_r)
+    x2 = min(WIDTH, cx + roi_r)
+    y1 = max(0, cy - roi_r)
+    y2 = min(HEIGHT, cy + roi_r)
+    roi = img_cv[y1:y2, x1:x2]
+    if roi.size == 0:
+        return None
+    r = roi[:, :, 0].astype(np.int32)
+    g = roi[:, :, 1].astype(np.int32)
+    b = roi[:, :, 2].astype(np.int32)
+    mask = (r > th) & (r > g * ratio) & (r > b * ratio)
+    oe = (r > 200) & (g > 200) & (b > 200) & (r >= g) & (r >= b) & ((r - g) > 10) & ((r - b) > 10)
+    combined = (mask | oe).astype(np.uint8) * 255
+    contours, _ = cv2.findContours(combined, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+    if not contours:
+        return None
+    largest = max(contours, key=cv2.contourArea)
+    if cv2.contourArea(largest) < 5:
+        return None
+    cnt = largest.copy()
+    for pt in cnt:
+        pt[0][0] += x1
+        pt[0][1] += y1
+    if len(cnt) >= 5:
+        (ex, ey), (ew, eh), ang = cv2.fitEllipse(cnt)
+        mask_ellipse = np.zeros((HEIGHT, WIDTH), dtype=np.uint8)
+        cv2.ellipse(mask_ellipse, (int(ex), int(ey)), (int(ew / 2), int(eh / 2)), ang, 0, 360, 255, -1)
+        brightness = img_cv[:, :, 0].astype(np.int32) + img_cv[:, :, 1].astype(np.int32) + img_cv[:, :, 2].astype(np.int32)
+        masked = np.where(mask_ellipse > 0, brightness, 0)
+        vals = masked[masked > 0]
+        if len(vals) > 0:
+            bth = np.percentile(vals, 90)
+            bmask = (masked >= bth).astype(np.uint8) * 255
+            bcontours, _ = cv2.findContours(bmask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+            if bcontours:
+                blargest = max(bcontours, key=cv2.contourArea)
+                if cv2.contourArea(blargest) >= 3 and len(blargest) >= 5:
+                    (ix, iy), _, _ = cv2.fitEllipse(blargest)
+                    return (float(ix), float(iy))
+                M = cv2.moments(blargest)
+                if M["m00"] > 0:
+                    return (float(M["m10"] / M["m00"]), float(M["m01"] / M["m00"]))
+        return (float(ex), float(ey))
+    M = cv2.moments(cnt)
+    if M["m00"] > 0:
+        return (float(M["m10"] / M["m00"]), float(M["m01"] / M["m00"]))
+    return None
+
+
+def find_brightest_bytes(frame, cx, cy, roi_r, th, ratio):
+    """使用 frame.to_bytes() 两阶段搜索，避免 cv2 转换"""
+    x1 = max(0, cx - roi_r)
+    x2 = min(WIDTH, cx + roi_r)
+    y1 = max(0, cy - roi_r)
+    y2 = min(HEIGHT, cy + roi_r)
+    data = frame.to_bytes()
+    best_score = 0
+    best_pos = None
+    # 第一阶段：隔点粗搜
+    for y in range(y1, y2, 2):
+        for x in range(x1, x2, 2):
+            idx = (y * WIDTH + x) * 3
+            r = data[idx]; g = data[idx+1]; b = data[idx+2]
+            if (r > th and r > g * ratio and r > b * ratio) or \
+               (r > 200 and g > 200 and b > 200 and r >= g and r >= b and (r - g) > 10 and (r - b) > 10):
+                score = r + g + b
+                dx = x - cx; dy = y - cy
+                score *= max(0.5, 1.0 - ((dx*dx + dy*dy) ** 0.5 / roi_r) * 0.5)
+                if score > best_score:
+                    best_score = score
+                    best_pos = (x, y)
+    if best_pos is None:
+        return None
+    # 第二阶段：候选点 7x7 精细搜索
+    fx, fy = best_pos
+    x1f = max(0, fx - 3); x2f = min(WIDTH, fx + 4)
+    y1f = max(0, fy - 3); y2f = min(HEIGHT, fy + 4)
+    best_bright = 0
+    final_pos = best_pos
+    for y in range(y1f, y2f):
+        for x in range(x1f, x2f):
+            idx = (y * WIDTH + x) * 3
+            r = data[idx]; g = data[idx+1]; b = data[idx+2]
+            if (r > th and r > g * ratio and r > b * ratio) or \
+               (r > 200 and g > 200 and b > 200 and r >= g and r >= b and (r - g) > 10 and (r - b) > 10):
+                rgb_sum = r + g + b
+                if rgb_sum > best_bright:
+                    best_bright = rgb_sum
+                    final_pos = (float(x), float(y))
+    return final_pos
+
+
+# 打开键盘输入设备
+_key_fd = None
+try:
+    _key_fd = open("/dev/input/event0", "rb")
+except Exception:
+    try:
+        _key_fd = open("/dev/input/event1", "rb")
+    except Exception:
+        _key_fd = None
+
+print("=" * 50)
+print("激光中心点检测单元测试")
+print("=" * 50)
+print()
+
+cam = camera.Camera(WIDTH, HEIGHT)
+disp = display.Display()
+print("[OK] 摄像头和显示初始化完成")
+
+# 初始化激光串口
+_laser_on = False
+_laser_uart = None
+try:
+    pinmap.set_pin_function("A18", "UART1_RX")
+    pinmap.set_pin_function("A19", "UART1_TX")
+    _laser_uart = uart.UART("/dev/ttyS1", 9600)
+    _laser_uart.read(-1)
+    print("[OK] 激光串口初始化完成")
+except Exception as e:
+    print(f"[WARN] 激光串口初始化失败: {e}")
+
+LASER_ON = bytes([0xAA, 0x00, 0x01, 0xBE, 0x00, 0x01, 0x00, 0x01, 0xC1])
+LASER_OFF = bytes([0xAA, 0x00, 0x01, 0xBE, 0x00, 0x01, 0x00, 0x00, 0xC0])
+
+# 默认开启激光
+if _laser_uart:
+    try:
+        _laser_uart.write(LASER_ON)
+        time.sleep_ms(50)
+        _laser_uart.read(-1)
+        _laser_on = True
+        print("[OK] 激光已开启")
+    except Exception as e:
+        print(f"[WARN] 开启激光失败: {e}")
+print()
+
+pos_ellipse = None
+pos_bright = None
+frame_count = 0
+use_ellipse = True
+
+while not app.need_exit():
+    frame = cam.read()
+    if frame is None:
+        time.sleep_ms(10)
+        continue
+
+    frame_count += 1
+
+    cx, cy = WIDTH // 2, HEIGHT // 2
+
+    t0 = time.ticks_ms()
+    pos_bright = find_brightest_bytes(frame, cx, cy, SEARCH_RADIUS, THRESHOLD, RED_RATIO)
+    t1 = time.ticks_ms()
+
+    pos_ellipse = None
+    if _USE_CV:
+        img_cv = image.image2cv(frame, False, False)
+        t2 = time.ticks_ms()
+        pos_ellipse = find_ellipse(img_cv, cx, cy, SEARCH_RADIUS, THRESHOLD, RED_RATIO)
+        t3 = time.ticks_ms()
+    else:
+        img_cv = None
+        t3 = t2 = t1
+
+    dt_b = abs(time.ticks_diff(t0, t1))
+    dt_e = abs(time.ticks_diff(t2, t3))
+
+    if frame_count % 5 == 0:
+        e_str = f"({pos_ellipse[0]:.1f},{pos_ellipse[1]:.1f})" if pos_ellipse else "None"
+        b_str = f"({pos_bright[0]:.1f},{pos_bright[1]:.1f})" if pos_bright else "None"
+        print(f"[LASER] ellipse={e_str} ({dt_e}ms)  brightest={b_str} ({dt_b}ms)  "
+              f"th={THRESHOLD}  ratio={RED_RATIO}  radius={SEARCH_RADIUS}")
+
+    pos = pos_ellipse if use_ellipse else pos_bright
+    if img_cv is not None:
+        cv2.circle(img_cv, (cx, cy), SEARCH_RADIUS, (0, 255, 0), 1)
+        cv2.circle(img_cv, (cx, cy), 2, (0, 255, 0), -1)
+        if pos:
+            x, y = int(pos[0]), int(pos[1])
+            cv2.circle(img_cv, (x, y), 6, (0, 0, 255), 2)
+            cv2.line(img_cv, (x - 14, y), (x + 14, y), (0, 0, 255), 1)
+            cv2.line(img_cv, (x, y - 14), (x, y + 14), (0, 0, 255), 1)
+            cv2.putText(img_cv, f"({x},{y})", (x + 10, y - 10),
+                        cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 1, cv2.LINE_AA)
+        info = [
+            f"pos={pos if pos else 'None'}",
+            f"method={'ellipse' if use_ellipse else 'brightest'}  th={THRESHOLD}  ratio={RED_RATIO}",
+            f"laser={'ON' if _laser_on else 'OFF'}",
+        ]
+        for i, line in enumerate(info):
+            cv2.putText(img_cv, line, (8, 20 + i * 22),
+                        cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 1, cv2.LINE_AA)
+        display_frame = image.cv2image(img_cv, False, False)
+    else:
+        display_frame = frame
+
+    disp.show(display_frame)
+
+    # 按键处理（非阻塞）
+    key = read_key_ev()
+    if key > 0:
+        c = chr(key & 0xFF) if key < 256 else ""
+        if key == 113 or key == 81 or key == 0x1b:  # q/Q/ESC
+            break
+        if c == "e" or key == 18:  # e
+            use_ellipse = not use_ellipse
+            print(f"[KEY] Method: {'ellipse' if use_ellipse else 'brightest'}")
+        if c == "l" or key == 12:  # l
+            _laser_on = not _laser_on
+            if _laser_uart:
+                try:
+                    _laser_uart.write(LASER_ON if _laser_on else LASER_OFF)
+                    time.sleep_ms(30)
+                    _laser_uart.read(-1)
+                    print(f"[KEY] Laser: {'ON' if _laser_on else 'OFF'}")
+                except Exception as e:
+                    print(f"[KEY] Laser error: {e}")
+            else:
+                print("[KEY] Laser UART not available")
+    time.sleep_ms(30)
+
+# 关闭激光
+if _laser_on and _laser_uart:
+    try:
+        _laser_uart.write(LASER_OFF)
+        _laser_uart.read(-1)
+        print("[EXIT] 激光已关闭")
+    except Exception:
+        pass
+print("[EXIT] 测试结束")
+if _key_fd:
+    _key_fd.close()

test/test_yolo26.py

@@ -0,0 +1,18 @@
+from maix import camera, display, image, nn, app
+
+# 1. 初始化模型 (请确保模型文件 .mud 路径正确)
+detector = nn.YOLOv5(model="/root/model_279350.mud", dual_buff=True)
+
+# 2. 初始化摄像头，分辨率与模型输入匹配
+cam = camera.Camera(detector.input_width(), detector.input_height(), detector.input_format())
+disp = display.Display()
+
+# 3. 主循环：实时检测与显示
+while not app.need_exit():
+    img = cam.read()                         # 从摄像头读取一帧
+    objs = detector.detect(img, conf_th=0.5, iou_th=0.45) # 执行YOLO11推理
+    for obj in objs:                         # 绘制所有检测到的目标
+        img.draw_rect(obj.x, obj.y, obj.w, obj.h, color=image.COLOR_RED)
+        msg = f'{detector.labels[obj.class_id]}: {obj.score:.2f}'
+        img.draw_string(obj.x, obj.y, msg, color=image.COLOR_RED)
+    disp.show(img)                           # 更新屏幕显示

test/test_yolo_camera_simple.py

@@ -0,0 +1,209 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+摄像头实时 YOLOv5 简易测试脚本。
+
+特点：
+- 完全独立脚本，直接 python test/test_yolo_camera_simple.py 运行，不需要传参。
+- 不 import config，不依赖项目模块。
+- 直接调用 maix.nn.YOLOv5(model=..., dual_buff=False)。
+- camera.read() 得到的 Maix image 直接送 det.detect()。
+- 在画面上画检测框、类别、置信度，并显示到屏幕。
+
+运行环境：MaixCAM / MaixPy。
+"""
+
+import os
+
+
+CAMERA_WIDTH = 640
+CAMERA_HEIGHT = 480
+# 默认与主项目 config.TRIANGLE_YOLO_MODEL_PATH 一致（勿用 /root/yolo26_int8.mud，那是占位路径）
+_MODEL_DEFAULT = "/maixapp/apps/t11/model_270139.mud"
+try:
+    import config as _cfg
+
+    MODEL_PATH = getattr(_cfg, "TRIANGLE_YOLO_MODEL_PATH", _MODEL_DEFAULT) or _MODEL_DEFAULT
+except Exception:
+    MODEL_PATH = _MODEL_DEFAULT
+CONF_TH = 0.7
+IOU_TH = 0.45
+# native: Maix detect 返回框已映射到 camera.read() 图像坐标；letterbox: 需要从网络输入坐标反算
+COORD_MODE = "native"
+# 只用于 DRAW_ONLY_CLASS_IDS=True 时过滤显示；默认画所有框
+CLASS_IDS = (0,)
+DRAW_ONLY_CLASS_IDS = False  # True=只画 CLASS_IDS 里的类别；False=画所有 YOLO 返回框
+
+
+def _det_obj_class_id(o):
+    for key in ("class_id", "cls", "label", "category", "cat_id", "id"):
+        if hasattr(o, key):
+            v = getattr(o, key)
+            if v is None:
+                continue
+            try:
+                return int(float(v))
+            except (TypeError, ValueError):
+                continue
+    return None
+
+
+def _det_obj_from_seq(t):
+    if not isinstance(t, (list, tuple)) or len(t) < 6:
+        return None
+
+    class Box:
+        pass
+
+    b = Box()
+    b.x = float(t[0])
+    b.y = float(t[1])
+    b.w = float(t[2])
+    b.h = float(t[3])
+    b.score = float(t[4])
+    b.class_id = int(float(t[5]))
+    return b
+
+
+def _normalize_objs(objs):
+    out = []
+    for o in objs or []:
+        if isinstance(o, (list, tuple)):
+            m = _det_obj_from_seq(o)
+            if m is not None:
+                out.append(m)
+        else:
+            out.append(o)
+    return out
+
+
+def _letterbox_net_to_src_xyxy(x, y, w, h, src_w, src_h, net_w, net_h):
+    scale = min(net_w / float(src_w), net_h / float(src_h))
+    new_w = src_w * scale
+    new_h = src_h * scale
+    pad_x = (net_w - new_w) * 0.5
+    pad_y = (net_h - new_h) * 0.5
+    x0 = (x - pad_x) / scale
+    y0 = (y - pad_y) / scale
+    x1 = (x + w - pad_x) / scale
+    y1 = (y + h - pad_y) / scale
+    return x0, y0, x1, y1
+
+
+def _det_to_src_xyxy(o, coord_mode, src_w, src_h, net_w, net_h):
+    x = float(getattr(o, "x", 0.0))
+    y = float(getattr(o, "y", 0.0))
+    w = float(getattr(o, "w", 0.0))
+    h = float(getattr(o, "h", 0.0))
+    if coord_mode in ("native", "source", "camera", "full"):
+        return x, y, x + w, y + h
+    return _letterbox_net_to_src_xyxy(x, y, w, h, src_w, src_h, net_w, net_h)
+
+
+def _clip_xywh(x0, y0, x1, y1, src_w, src_h):
+    x0 = max(0, min(int(round(x0)), src_w - 1))
+    y0 = max(0, min(int(round(y0)), src_h - 1))
+    x1 = max(x0 + 1, min(int(round(x1)), src_w))
+    y1 = max(y0 + 1, min(int(round(y1)), src_h))
+    return x0, y0, x1 - x0, y1 - y0
+
+
+def _label(det, cid):
+    labels = getattr(det, "labels", None)
+    if labels is None:
+        return str(cid)
+    try:
+        return str(labels[int(cid)])
+    except Exception:
+        return str(cid)
+
+
+def main():
+    from maix import camera, display, nn, time, image
+
+    if not MODEL_PATH or not os.path.isfile(MODEL_PATH):
+        print("[ERR] 模型文件不存在:", MODEL_PATH)
+        return
+
+    print("[INFO] 初始化 YOLO 模型:", MODEL_PATH)
+    det = nn.YOLOv26(model=MODEL_PATH, dual_buff=False)
+    net_w = int(det.input_width())
+    net_h = int(det.input_height())
+    print(
+        "[INFO] net_in=%dx%d conf=%.2f iou=%.2f coord=%s class_ids=%s"
+        % (net_w, net_h, CONF_TH, IOU_TH, COORD_MODE, str(CLASS_IDS))
+    )
+
+    print("[INFO] 初始化摄像头: %dx%d" % (CAMERA_WIDTH, CAMERA_HEIGHT))
+    cam = camera.Camera(CAMERA_WIDTH, CAMERA_HEIGHT)
+    disp = display.Display()
+
+    color_cycle = []
+    for name in ("RED", "GREEN", "BLUE", "ORANGE", "YELLOW", "CYAN", "MAGENTA"):
+        c = getattr(image, "COLOR_" + name, None)
+        if c is not None:
+            color_cycle.append(c)
+    if not color_cycle:
+        color_cycle = [getattr(image, "COLOR_RED", 0)]
+
+    frame_idx = 0
+    last_log_ms = time.ticks_ms()
+    fps_count = 0
+
+    while True:
+        frame = cam.read()
+        src_w = frame.width()
+        src_h = frame.height()
+
+        t0 = time.ticks_ms()
+        raw = det.detect(frame, conf_th=CONF_TH, iou_th=IOU_TH)
+        detect_ms = time.ticks_ms() - t0
+        objs = _normalize_objs(raw if raw is not None else [])
+
+        draw_count = 0
+        for i, o in enumerate(objs):
+            cid = _det_obj_class_id(o)
+            if cid is None:
+                cid = -1
+            if DRAW_ONLY_CLASS_IDS and cid not in CLASS_IDS:
+                continue
+            try:
+                score = float(getattr(o, "score", 0.0))
+            except Exception:
+                score = 0.0
+
+            x0, y0, x1, y1 = _det_to_src_xyxy(o, COORD_MODE, src_w, src_h, net_w, net_h)
+            ix, iy, iw, ih = _clip_xywh(x0, y0, x1, y1, src_w, src_h)
+            col = color_cycle[cid % len(color_cycle)] if cid >= 0 else color_cycle[0]
+            frame.draw_rect(ix, iy, iw, ih, color=col)
+            frame.draw_string(ix, max(0, iy - 16), "%s %.2f" % (_label(det, cid), score), color=col)
+            draw_count += 1
+
+        frame.draw_string(4, 4, "YOLO boxes:%d draw:%d %dms" % (len(objs), draw_count, detect_ms), color=color_cycle[0])
+        disp.show(frame)
+
+        frame_idx += 1
+        fps_count += 1
+        now = time.ticks_ms()
+        if now - last_log_ms >= 1000:
+            print(
+                "[INFO] frame=%d fps=%d raw_boxes=%d draw_boxes=%d detect_ms=%d"
+                % (frame_idx, fps_count, len(objs), draw_count, detect_ms)
+            )
+            fps_count = 0
+            last_log_ms = now
+
+
+if __name__ == "__main__":
+    try:
+        main()
+    except KeyboardInterrupt:
+        print("[INFO] exit")
+    except Exception as e:
+        print("[ERR]", e)
+        try:
+            import traceback
+
+            traceback.print_exc()
+        except Exception:
+            pass

test/test_yolov8.py

@@ -0,0 +1,29 @@
+from maix import image, nn, display
+
+# 1. 加载模型
+detector = nn.YOLOv8(model="/root/279350.mud", dual_buff=False)
+# 2. 加载指定图片（根据模型输入尺寸自动缩放宽高）
+img = image.load("/root/tes.jpg")
+if img is None:
+    raise FileNotFoundError("图片加载失败，请检查路径")
+
+# 3. 调整图片尺寸到模型输入要求（可选，detect内部会处理，但提前缩放可提高速度）
+# img = img.resize(detector.input_width(), detector.input_height())
+
+# 4. 检测
+objs = detector.detect(img, conf_th=0.5, iou_th=0.45)
+
+# 5. 在图片上绘制结果
+for obj in objs:
+    img.draw_rect(obj.x, obj.y, obj.w, obj.h, color=image.COLOR_RED)
+    msg = f'{detector.labels[obj.class_id]}: {obj.score:.2f}'
+    img.draw_string(obj.x, obj.y, msg, color=image.COLOR_RED)
+
+# 6. 显示结果（如果设备有屏幕）
+disp = display.Display()
+disp.show(img)
+
+# 7. 保存结果（可选）
+img.save("/root/result.jpg")
+
+print("识别完成，结果已显示并保存为 result.jpg")

triangle_target.py

@@ -22,6 +22,143 @@ def _log(msg):
         pass
 
 
+def _read_triangle_direction_cfg():
+    """读取 config 中三角形方向/中心距校验参数。"""
+    try:
+        import config as cfg
+        return {
+            "enable": bool(getattr(cfg, "TRIANGLE_DIRECTION_VALIDATE_ENABLE", True)),
+            "min_pass": int(getattr(cfg, "TRIANGLE_DIRECTION_MIN_PASS", 3)),
+            "dot_min": float(getattr(cfg, "TRIANGLE_DIRECTION_DOT_MIN", 0.0)),
+            "to_center_dot_min": float(
+                getattr(cfg, "TRIANGLE_DIRECTION_TO_CENTER_DOT_MIN", 0.35)
+            ),
+            "center_dist_enable": bool(
+                getattr(cfg, "TRIANGLE_CENTER_DISTANCE_VALIDATE_ENABLE", True)
+            ),
+            "center_dist_tol": float(
+                getattr(cfg, "TRIANGLE_CENTER_DISTANCE_RATIO_TOL", 0.45)
+            ),
+        }
+    except Exception:
+        return {
+            "enable": True,
+            "min_pass": 3,
+            "dot_min": 0.0,
+            "to_center_dot_min": 0.35,
+            "center_dist_enable": True,
+            "center_dist_tol": 0.45,
+        }
+
+
+def _quad_combo_orient_penalty(cands_4):
+    """
+    四点组合评分用的方向惩罚（原 _score_quad 内 orient_pen 逻辑）。
+    TRIANGLE_DIRECTION_VALIDATE_ENABLE=False 时调用方应跳过（不加罚）。
+    """
+    orient_pen = 0.0
+    orient_vote = []
+    for c in cands_4:
+        cen = np.array(c["center_px"], dtype=np.float32)
+        rpt = np.array(c["right_pt"], dtype=np.float32)
+        vx = float(cen[0] - rpt[0])
+        vy = float(cen[1] - rpt[1])
+        if abs(vx) < 1e-6 or abs(vy) < 1e-6:
+            orient_pen += 1.0
+            orient_vote.append(None)
+            continue
+        if abs(vx) < abs(vy) * 0.15 or abs(vy) < abs(vx) * 0.15:
+            orient_pen += 0.5
+        if vx > 0 and vy > 0:
+            orient_vote.append(0)
+        elif vx < 0 and vy > 0:
+            orient_vote.append(1)
+        elif vx > 0 and vy < 0:
+            orient_vote.append(2)
+        else:
+            orient_vote.append(3)
+    valid_votes = [v for v in orient_vote if v is not None]
+    if valid_votes:
+        from collections import Counter
+        vc = Counter(valid_votes)
+        orient_pen += max(0, max(vc.values()) - 1) * 0.8
+    return orient_pen
+
+
+def _marker_inward_unit(marker):
+    """从直角顶点指向三角内部的单位向量；marker['center'] 为直角顶点。"""
+    right = np.array(marker["center"], dtype=np.float64)
+    corners = marker.get("corners")
+    if not corners or len(corners) < 3:
+        return None
+    cen = np.mean(np.array(corners, dtype=np.float64), axis=0)
+    inv = cen - right
+    n = float(np.linalg.norm(inv))
+    if n < 1e-6:
+        return None
+    return inv / n
+
+
+def _validate_triangle_direction(marker_centers, tri_markers, cfg):
+    """
+    校验：四角到候选靶心距离近似一致；各真实黑三角朝向靶心。
+    仅统计 tri_markers 中真实检出的角（不含几何补全的虚拟点）。
+    Returns:
+        (ok: bool, reason: str)
+    """
+    if not cfg.get("enable", True):
+        return True, ""
+
+    pts = np.array(marker_centers, dtype=np.float64).reshape(-1, 2)
+    if len(pts) < 3:
+        return True, ""
+
+    quad_center = np.mean(pts, axis=0)
+
+    if cfg.get("center_dist_enable", True) and len(pts) >= 3:
+        dists = np.linalg.norm(pts - quad_center, axis=1)
+        mean_d = float(np.mean(dists))
+        if mean_d > 1e-6:
+            ratio = (float(np.max(dists)) - float(np.min(dists))) / mean_d
+            tol = float(cfg.get("center_dist_tol", 0.45))
+            if ratio > tol:
+                return False, f"center_dist_ratio={ratio:.2f}>{tol:.2f}"
+
+    dot_need = max(
+        float(cfg.get("dot_min", 0.0)),
+        float(cfg.get("to_center_dot_min", 0.35)),
+    )
+    pass_n = 0
+    check_n = 0
+    for m in tri_markers or []:
+        if m.get("center") is None:
+            continue
+        check_n += 1
+        right = np.array(m["center"], dtype=np.float64)
+        to_center = quad_center - right
+        nc = float(np.linalg.norm(to_center))
+        if nc < 1e-6:
+            continue
+        inward = _marker_inward_unit(m)
+        if inward is None:
+            continue
+        dot_tc = float(np.dot(inward, to_center / nc))
+        if dot_tc >= dot_need:
+            pass_n += 1
+
+    if check_n == 0:
+        return True, ""
+
+    min_pass = int(cfg.get("min_pass", 3))
+    min_pass = max(1, min(min_pass, check_n))
+    if pass_n < min_pass:
+        return False, (
+            f"direction_pass={pass_n}/{check_n} need>={min_pass} "
+            f"(dot>={dot_need:.2f})"
+        )
+    return True, ""
+
+
 def _gray_suppress_bright_by_v(img_rgb, v_above: int):
     """
     RGB 输入：在 HSV 的 V 上，将亮度 >= v_above 的像素灰度置为 255。
@@ -224,7 +361,7 @@ def detect_triangle_markers(
         blackhat_kernel_frac = 0.018
     try:
         import config as _tcfg
-        _timing_log = bool(getattr(_tcfg, "TRIANGLE_TIMING_LOG", True))
+        _timing_log = bool(getattr(_tcfg, "ARCHERY_TIMING_ENABLE", True)) and bool(getattr(_tcfg, "TRIANGLE_TIMING_LOG", True))
     except Exception:
         _timing_log = True
 
@@ -622,6 +759,8 @@ def detect_triangle_markers(
         bot_pair = sorted(by_y[2:], key=lambda i: pts_4[i][0])
         return top_pair[0], bot_pair[0], bot_pair[1], top_pair[1]
 
+    _dir_cfg_combo = _read_triangle_direction_cfg()
+
     def _score_quad(cands_4):
         pts = [np.array(c["center_px"]) for c in cands_4]
         legs = [c["avg_leg"] for c in cands_4]
@@ -641,7 +780,13 @@ def detect_triangle_markers(
         med_l = float(np.median(legs))
         leg_dev = max(abs(l - med_l) / (med_l + 1e-6) for l in legs)
 
-        score = (diag_ratio - 1.0) * 3.0 + (h_ratio - 1.0) + (v_ratio - 1.0) + leg_dev * 2.0
+        orient_pen = (
+            _quad_combo_orient_penalty(cands_4)
+            if _dir_cfg_combo.get("enable", True)
+            else 0.0
+        )
+
+        score = (diag_ratio - 1.0) * 3.0 + (h_ratio - 1.0) + (v_ratio - 1.0) + leg_dev * 2.0 + orient_pen
         return score, (tl, bl, br, tr)
 
     assigned = None
@@ -932,6 +1077,8 @@ def _assign_marker_ids_from_filtered(filtered, verbose=True):
         bot_pair = sorted(by_y[2:], key=lambda i: pts_4[i][0])
         return top_pair[0], bot_pair[0], bot_pair[1], top_pair[1]
 
+    _dir_cfg_combo = _read_triangle_direction_cfg()
+
     def _score_quad(cands_4):
         pts = [np.array(c["center_px"]) for c in cands_4]
         legs = [c["avg_leg"] for c in cands_4]
@@ -947,7 +1094,12 @@ def _assign_marker_ids_from_filtered(filtered, verbose=True):
         v_ratio = max(s_left, s_right) / (min(s_left, s_right) + 1e-6)
         med_l = float(np.median(legs))
         leg_dev = max(abs(l - med_l) / (med_l + 1e-6) for l in legs)
-        score = (diag_ratio - 1.0) * 3.0 + (h_ratio - 1.0) + (v_ratio - 1.0) + leg_dev * 2.0
+        orient_pen = (
+            _quad_combo_orient_penalty(cands_4)
+            if _dir_cfg_combo.get("enable", True)
+            else 0.0
+        )
+        score = (diag_ratio - 1.0) * 3.0 + (h_ratio - 1.0) + (v_ratio - 1.0) + leg_dev * 2.0 + orient_pen
         return score, (tl, bl, br, tr)
 
     assigned = None
@@ -1113,7 +1265,7 @@ def try_triangle_scoring(
 
     try:
         import config as _cfg_tl
-        _try_timing_log = bool(getattr(_cfg_tl, "TRIANGLE_TIMING_LOG", True))
+        _try_timing_log = bool(getattr(_cfg_tl, "ARCHERY_TIMING_ENABLE", True)) and bool(getattr(_cfg_tl, "TRIANGLE_TIMING_LOG", True))
         _crop_min_side = int(getattr(_cfg_tl, "TRIANGLE_CROP_ROI_MIN_SIDE_PX", 64))
     except Exception:
         _try_timing_log = True
@@ -1733,6 +1885,21 @@ def try_triangle_scoring(
             "is_virtual": bool(_is_virtual),
         })
 
+    # ---------- 方向 / 中心距校验（config.TRIANGLE_DIRECTION_*） ----------
+    _dir_cfg = _read_triangle_direction_cfg()
+    _dir_ok, _dir_reason = _validate_triangle_direction(
+        marker_centers, tri_markers, _dir_cfg
+    )
+    if not _dir_ok:
+        _log(f"[TRI] 方向校验失败: {_dir_reason}")
+        if _try_timing_log:
+            _log(
+                f"[TRI] timing_ms(try_triangle): {_tri_yolo_part} "
+                f"geometry={(time.perf_counter() - _t_seg) * 1000:.1f} "
+                f"total_try={(time.perf_counter() - _t_try0) * 1000:.1f} (方向校验失败)"
+            )
+        return out
+
     # ---------- 结果有效性校验（防 nan/inf 与退化角点） ----------
     try:
         import config as _cfg

version.md

@@ -1,28 +0,0 @@
-# 1.2.0 开始使用C++编译成.so，替换部分代码
-# 1.2.1 ota使用加密包
-# 1.2.2 支持wifi ota，并且设定时区，并使用单独线程保存图片
-# 1.2.3 修改ADC_TRIGGER_THRESHOLD 为2300，支持上传日志到服务器
-# 1.2.4 修改ADC_TRIGGER_THRESHOLD 为3000，并默认关闭摄像头的显示，并把ADC的采样间隔从50ms降低到10ms
-# 1.2.5 支持空气传感器采样，并默认关闭日志。优化断网时的发送队列丢消息问题，解决 WiFi 断线检测不可靠问题。
-# 1.2.6 在链接 wifi 前先判断 wifi 的可用性，假如不可用，则不落盘。增加日志批量压缩上传功能
-# 1.2.7 修复OTA失败的bug, 空气压力传感器的阈值是2500
-# 1.2.8 （1） 加快 wifi 下数据传输的速度。（2） 调整射箭时处理的逻辑，优先上报数据，再存照片之类的操作。（3）假如是用户打开激光的，射箭触发后不再关闭激光，因为是调瞄阶段
-# 1.2.9 增加电源板的控制和自动关机的功能
-# 1.2.10 config formal
-# 1.2.11 增加三角形的单应性算法，适配对应的靶纸
-# 1.2.110 关掉了黑色三角形算法，只用于测试
-# 1.2.13 修改wifi连接
-# 1.2.14 修改了icc登录部分
-# 2.15.3 新版本ota，去除ai算环数方法
-# 2.15.4 更新版本号
-# 2.15.5 打印ota进度
-# 2.15.6 更新版本号
-# 2.15.7 更新版本号
-# 2.15.8 启动不加载预加载yolo
-# 2.15.9 20cm
-# 2.15.10 不保存图片
-# 2.15.11 优化内存
-# 2.15.12 优化算法
-# 2.15.13 优化算法
-# 2.15.14 优化算法
-# 2.15.15 优化wifi连接

version.py

@@ -4,6 +4,28 @@
 应用版本号
 每次 OTA 更新时，只需要更新这个文件中的版本号
 """
-VERSION = '2.15.15'
+VERSION = '1.2.15.1'
+
+
+# 1.2.0 开始使用C++编译成.so，替换部分代码
+# 1.2.1 ota使用加密包
+# 1.2.2 支持wifi ota，并且设定时区，并使用单独线程保存图片
+# 1.2.3 修改ADC_TRIGGER_THRESHOLD 为2300，支持上传日志到服务器
+# 1.2.4 修改ADC_TRIGGER_THRESHOLD 为3000，并默认关闭摄像头的显示，并把ADC的采样间隔从50ms降低到10ms
+# 1.2.5 支持空气传感器采样，并默认关闭日志。优化断网时的发送队列丢消息问题，解决 WiFi 断线检测不可靠问题。
+# 1.2.6 在链接 wifi 前先判断 wifi 的可用性，假如不可用，则不落盘。增加日志批量压缩上传功能
+# 1.2.7 修复OTA失败的bug, 空气压力传感器的阈值是2500
+# 1.2.8 （1） 加快 wifi 下数据传输的速度。（2） 调整射箭时处理的逻辑，优先上报数据，再存照片之类的操作。（3）假如是用户打开激光的，射箭触发后不再关闭激光，因为是调瞄阶段
+# 1.2.9 增加电源板的控制和自动关机的功能
+# 1.2.10 config formal
+# 1.2.11 增加三角形的单应性算法，适配对应的靶纸
+# 1.2.110 关掉了黑色三角形算法，只用于测试
+# 1.2.13 修改wifi连接
+# 1.2.14 修改了icc登录部分
+# 1.2.15.1 增加了标靶判断 20 40 
+# 1.2.16.1 增加激光校准，三角形方向判断，时间开关
+
+
+
 
 

vision.py

@@ -10,6 +10,7 @@ import os
 import math
 import threading
 import queue
+import time
 from maix import image
 import config
 from logger_manager import logger_manager
@@ -531,11 +532,14 @@ def detect_circle_v3(frame, laser_point=None, img_cv=None):
     if img_cv is None:
         img_cv = image.image2cv(frame, False, False)
     logger = logger_manager.logger
+    _timing_on = bool(getattr(config, "VISION_TIMING_ENABLE", True))
+    _t0 = time.perf_counter() if _timing_on else None
+    _t1 = _t2 = _t3 = _t4 = _t5 = None
     from datetime import datetime
     logger.debug(f"[detect_circle_v3] begin {datetime.now()}")
     # -- 1. 缩图加速（与三角形路径保持一致）
     h_orig, w_orig = img_cv.shape[:2]
-    MAX_DET_DIM = 480
+    MAX_DET_DIM = 320
     long_side = max(h_orig, w_orig)
     if long_side > MAX_DET_DIM:
         det_scale = MAX_DET_DIM / long_side
@@ -554,6 +558,8 @@ def detect_circle_v3(frame, laser_point=None, img_cv=None):
     ellipse_params = None
     
     logger.debug(f"[detect_circle_v3] step 1 fin {datetime.now()}")
+    if _timing_on:
+        _t1 = time.perf_counter()
 
     # -- 2. HSV + 黄色掩码
     hsv = cv2.cvtColor(img_det, cv2.COLOR_RGB2HSV)
@@ -567,25 +573,26 @@ def detect_circle_v3(frame, laser_point=None, img_cv=None):
     mask_yellow = cv2.morphologyEx(mask_yellow, cv2.MORPH_CLOSE, kernel)
 
     logger.debug(f"[detect_circle_v3] step 2 fin {datetime.now()}")
+    if _timing_on:
+        _t2 = time.perf_counter()
+        _t3 = time.perf_counter()
 
     # -- 3. 红色掩码：在循环外只算一次
     mask_red = cv2.bitwise_or(
-        cv2.inRange(hsv, np.array([0, 30, 20]),   np.array([12,  255, 255])),
+        cv2.inRange(hsv, np.array([0, 80, 0]),   np.array([10,  255, 255])),
-        cv2.inRange(hsv, np.array([168, 30, 20]), np.array([180, 255, 255])),
+        cv2.inRange(hsv, np.array([170, 80, 0]), np.array([180, 255, 255])),
     )
     kernel_red = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
     mask_red = cv2.morphologyEx(mask_red, cv2.MORPH_CLOSE, kernel_red)
-    # 再加一次膨胀，加厚环状区域避免碎片化
-    mask_red = cv2.dilate(mask_red, kernel_red, iterations=1)
     contours_red, _ = cv2.findContours(mask_red, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
     # 预先把红色轮廓筛选成 (center, radius) 列表，后续直接查表
     red_candidates = []
     for cnt_r in contours_red:
         ar = cv2.contourArea(cnt_r)
-        if ar <= 10:
+        if ar <= 50:
             continue
         pr = cv2.arcLength(cnt_r, True)
-        if pr <= 0 or (4 * np.pi * ar) / (pr * pr) <= 0.2:
+        if pr <= 0 or (4 * np.pi * ar) / (pr * pr) <= 0.6:
             continue
         if len(cnt_r) >= 5:
             (xr, yr), (wr, hr), _ = cv2.fitEllipse(cnt_r)
@@ -595,19 +602,22 @@ def detect_circle_v3(frame, laser_point=None, img_cv=None):
             red_candidates.append({"center": (int(xr), int(yr)), "radius": int(rr)})
 
     logger.debug(f"[detect_circle_v3] step 3 fin {datetime.now()}")
+    if _timing_on:
+        _t3 = time.perf_counter()
+        _t4 = time.perf_counter()
 
     # -- 4. 黄色轮廓循环（复用上面的红色候选列表）
     contours_yellow, _ = cv2.findContours(mask_yellow, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
     valid_targets = []
     for cnt_yellow in contours_yellow:
         area = cv2.contourArea(cnt_yellow)
-        if area <= 15:
+        if area <= 50:
             continue
         perimeter = cv2.arcLength(cnt_yellow, True)
         if perimeter <= 0:
             continue
         circularity = (4 * np.pi * area) / (perimeter * perimeter)
-        if circularity <= 0.5:
+        if circularity <= 0.7:
             continue
         if logger:
             logger.info(f"[target] -> 面积:{area:.1f}, 圆度:{circularity:.2f}")
@@ -627,11 +637,7 @@ def detect_circle_v3(frame, laser_point=None, img_cv=None):
             ddx = yellow_center[0] - rc["center"][0]
             ddy = yellow_center[1] - rc["center"][1]
             dist_centers = math.hypot(ddx, ddy)
-            max_dist = yellow_radius * 2.0
+            if dist_centers < yellow_radius * 1.5 and rc["radius"] > yellow_radius * 0.8:
-            min_r = min(rc["radius"], yellow_radius)
-            max_r = max(rc["radius"], yellow_radius)
-            size_ratio = min_r / max_r if max_r > 0 else 0
-            if dist_centers < max_dist and size_ratio > 0.5:
                 if logger:
                     logger.info(f"[target] -> 找到匹配的红圈: 黄心({yellow_center}), "
                                 f"红心({rc['center']}), 距离:{dist_centers:.1f}, "
@@ -644,19 +650,13 @@ def detect_circle_v3(frame, laser_point=None, img_cv=None):
                 })
                 matched = True
                 break
-        if not matched:
+        if not matched and logger:
-            # 黄圈高置信度兜底：大且圆时跳过红圈验证
+            logger.debug("Debug -> 未找到匹配的红色圆圈，可能是误识别")
-            if area > 30 and circularity > 0.8:
-                valid_targets.append({
-                    "center": yellow_center,
-                    "radius": yellow_radius,
-                    "ellipse": yellow_ellipse,
-                    "area": area,
-                })
-            elif logger:
-                logger.debug("Debug -> 未找到匹配的红色圆圈，可能是误识别")
 
     logger.debug(f"[detect_circle_v3] step 4 fin {datetime.now()}")
+    if _timing_on:
+        _t4 = time.perf_counter()
+        _t5 = time.perf_counter()
 
     # -- 5. 选最佳目标，坐标还原到原始分辨率
     if valid_targets:
@@ -684,7 +684,20 @@ def detect_circle_v3(frame, laser_point=None, img_cv=None):
         ellipse_params = be
         best_radius1 = best_radius * 5
     result_img = image.cv2image(img_cv, False, False)
-    logger.debug(f"[detect_circle_v3] step 5 fin {datetime.now()}")
+    if _timing_on:
+        _t5 = time.perf_counter()
+        _t_all = (_t5 - _t0) * 1000
+        _ms1 = (_t1 - _t0) * 1000
+        _ms2 = (_t2 - _t1) * 1000
+        _ms3 = (_t3 - _t2) * 1000
+        _ms4 = (_t4 - _t3) * 1000
+        _ms5 = (_t5 - _t4) * 1000
+        logger.info(
+            f"[VISION timing] total={_t_all:.1f}ms  "
+            f"resize={_ms1:.1f}  hsv_yellow={_ms2:.1f}  "
+            f"red_mask={_ms3:.1f}  yellow_loop={_ms4:.1f}  "
+            f"select_cv2img={_ms5:.1f}"
+        )
     return result_img, best_center, best_radius, method, best_radius1, ellipse_params
 
 def estimate_distance(pixel_radius):
@@ -936,6 +949,51 @@ def start_save_shot_worker():
         logger.info("[VISION] 存图 worker 线程已启动")
 
 
+def enqueue_save_raw_shot(frame, shot_id=None, photo_dir=None):
+    """
+    异步保存射箭原图（无算法标注）。需 SAVE_IMAGE_ENABLED 且 SAVE_RAW_SHOT_IMAGE_ENABLED。
+    文件名：{photo_dir}/shot_{shot_id}_raw.jpg
+    """
+    if not getattr(config, "SAVE_RAW_SHOT_IMAGE_ENABLED", False):
+        return
+    if not getattr(config, "SAVE_IMAGE_ENABLED", True):
+        return
+    if not shot_id:
+        return
+    if photo_dir is None:
+        photo_dir = config.PHOTO_DIR
+
+    try:
+        img_cv = image.image2cv(frame, False, False)
+        img_copy = np.copy(img_cv)
+    except Exception as e:
+        logger = logger_manager.logger
+        if logger:
+            logger.error(f"[VISION] enqueue_save_raw_shot 复制图像失败: {e}")
+        return
+
+    def _job():
+        try:
+            try:
+                if photo_dir not in os.listdir("/root"):
+                    os.mkdir(photo_dir)
+            except Exception:
+                pass
+            filename = f"{photo_dir}/shot_{shot_id}_raw.jpg"
+            out = image.cv2image(img_copy, False, False)
+            out.save(filename)
+            logger = logger_manager.logger
+            if logger:
+                logger.info(f"[VISION] 已保存射箭原图: {filename}")
+            prune_old_images_in_dir(photo_dir, config.MAX_IMAGES, logger, "[VISION]")
+        except Exception as e:
+            logger = logger_manager.logger
+            if logger:
+                logger.error(f"[VISION] 保存射箭原图失败: {e}")
+
+    threading.Thread(target=_job, daemon=True).start()
+
+
 def enqueue_save_shot(result_img, center, radius, method, ellipse_params,
                       laser_point, distance_m, shot_id=None, photo_dir=None,
                       yolo_roi_xyxy=None):
@@ -1025,3 +1083,63 @@ def detect_target(frame, laser_point=None):
         logger.debug("[VISION] 使用传统黄色靶心检测")
     return detect_circle_v3(frame, laser_point)
 
+
+def sample_target_rgb_at_physical_radius(frame, target_center, target_radius_px, radius_cm=None, angles_deg=None, patch_half_px=None, black_thresh=None, timing=False):
+    """
+    在物方半径位置采样 RGB，判断黑/白靶。
+    返回: dict {ok, is_black, mean_rgb, samples, black_ratio, elapsed_ms}
+    """
+    logger = logger_manager.logger
+    if target_center is None or target_radius_px is None:
+        return {"ok": False, "reason": "no_target", "is_black": None, "elapsed_ms": 0.0}
+
+    radius_cm = float(radius_cm if radius_cm is not None else getattr(config, "TRIANGLE_SAMPLE_RADIUS_CM", 15.0))
+    angles_deg = tuple(angles_deg if angles_deg is not None else getattr(config, "TRIANGLE_SAMPLE_ANGLES_DEG", (0, 90, 180, 270)))
+    patch_half_px = int(patch_half_px if patch_half_px is not None else getattr(config, "TRIANGLE_SAMPLE_PATCH_HALF_PX", 2))
+    black_thresh = float(black_thresh if black_thresh is not None else getattr(config, "TRIANGLE_SAMPLE_BLACK_THRESH", 30.0))
+    timing_on = bool(timing) and bool(getattr(config, "TRIANGLE_SAMPLE_TIMING_ENABLE", True))
+    t0 = time.perf_counter() if timing_on else None
+
+    try:
+        img_cv = image.image2cv(frame, False, False)
+        h, w = img_cv.shape[:2]
+        cx, cy = float(target_center[0]), float(target_center[1])
+        scale = float(target_radius_px) / max(radius_cm, 1e-6)
+        samples = []
+        black_count = 0
+        for ang in angles_deg:
+            rad = math.radians(float(ang))
+            sx = int(round(cx + math.cos(rad) * radius_cm * scale))
+            sy = int(round(cy + math.sin(rad) * radius_cm * scale))
+            x0 = max(0, sx - patch_half_px)
+            y0 = max(0, sy - patch_half_px)
+            x1 = min(w, sx + patch_half_px + 1)
+            y1 = min(h, sy + patch_half_px + 1)
+            if x1 <= x0 or y1 <= y0:
+                continue
+            patch = img_cv[y0:y1, x0:x1]
+            mean_rgb = patch.reshape(-1, 3).mean(axis=0)
+            is_black = bool(np.all(mean_rgb < black_thresh))
+            black_count += 1 if is_black else 0
+            samples.append({"angle": float(ang), "xy": (sx, sy), "mean_rgb": tuple(float(v) for v in mean_rgb), "is_black": is_black})
+        black_ratio = float(black_count) / float(len(samples) or 1)
+        out = {
+            "ok": len(samples) > 0,
+            "is_black": black_ratio >= 0.5,
+            "mean_rgb": tuple(float(v) for v in (np.mean([s["mean_rgb"] for s in samples], axis=0) if samples else (0, 0, 0))),
+            "samples": samples,
+            "black_ratio": black_ratio,
+            "elapsed_ms": (time.perf_counter() - t0) * 1000.0 if timing_on else 0.0,
+        }
+        if logger:
+            logger.info(
+                f"[TRI-SAMPLE] radius_cm={radius_cm:.1f} black_thresh={black_thresh:.1f} "
+                f"black_ratio={black_ratio:.2f} is_black={out['is_black']} "
+                f"elapsed_ms={out['elapsed_ms']:.1f} samples={len(samples)}"
+            )
+        return out
+    except Exception as e:
+        if logger:
+            logger.error(f"[TRI-SAMPLE] 采样失败: {e}")
+        return {"ok": False, "reason": str(e), "is_black": None, "elapsed_ms": 0.0}
+

wifi.py

@@ -41,7 +41,6 @@ class WiFiManager:
         # WiFi 质量监测（后台线程）
         self._wifi_quality_monitor_thread = None
         self._wifi_quality_stop_event = threading.Event()
-        self._wifi_quality_lock = threading.Lock()
         self._last_wifi_rtt_ms = None  # 最近一次测量的 RTT
         self._last_wifi_rssi_dbm = None  # 最近一次测量的 RSSI
         
@@ -239,6 +238,7 @@ class WiFiManager:
         old_conf = _read_text(conf_path)
         old_boot_ssid = _read_text(ssid_file)
         old_boot_pass = _read_text(pass_file)
+        old_boot_wpa = _read_text(boot_wpa_path) if os.path.exists(boot_wpa_path) else None
 
         try:
             try:
@@ -250,13 +250,9 @@ class WiFiManager:
                 _write_text(conf_path, full_conf)
             except Exception:
                 pass
-            # 删除 wpa_supplicant.conf，让 S30wifi 回退读 ssid/pass
+            _write_text(boot_wpa_path, full_conf)
-            try:
-                if os.path.exists(boot_wpa_path):
-                    os.remove(boot_wpa_path)
-            except Exception:
-                pass
 
+            # 仍写入 ssid/pass，便于其它脚本/人工查看；S30wifi 优先使用 wpa_supplicant.conf
             _write_text(ssid_file, ssid.strip())
             _write_text(pass_file, password.strip())
 
@@ -296,6 +292,7 @@ class WiFiManager:
                     if not persist:
                         # 不持久化：把 /boot 恢复成旧值（不重启，当前连接保持不变）
                         _restore_boot(old_boot_ssid, old_boot_pass)
+                        _restore_boot_wpa(old_boot_wpa)
                         self.logger.info("[WIFI] 网络验证通过，但按 persist=False 回滚 /boot 凭证（不重启）")
                     else:
                         self.logger.info("[WIFI] 网络验证通过，/boot 凭证已保留（持久化）")
@@ -309,6 +306,7 @@ class WiFiManager:
         except Exception as e:
             # 失败：回滚 /boot 和 /etc，重启 WiFi 恢复旧网络
             _restore_boot(old_boot_ssid, old_boot_pass)
+            _restore_boot_wpa(old_boot_wpa)
             try:
                 if old_conf is not None:
                     _write_text(conf_path, old_conf)
@@ -353,11 +351,7 @@ class WiFiManager:
             else:
                 full_conf = build_sta_conf_open(ssid)
             _write_text(conf_path, full_conf)
-            try:
+            _write_text(boot_wpa_path, full_conf)
-                if os.path.exists(boot_wpa_path):
-                    os.remove(boot_wpa_path)
-            except Exception:
-                pass
         except ValueError as e:
             return False, str(e)
         except Exception as e:
@@ -548,45 +542,34 @@ class WiFiManager:
             network_type_callback: 获取当前网络类型的回调函数
             on_poor_quality_callback: WiFi质量差时的回调函数
         """
-        with self._wifi_quality_lock:
+        if self._wifi_quality_monitor_thread is not None:
-            if self._wifi_quality_monitor_thread is not None and self._wifi_quality_monitor_thread.is_alive():
+            self.logger.warning("[WiFi Monitor] 监测线程已在运行")
-                self.logger.warning("[WiFi Monitor] 监测线程已在运行")
+            return
-                return
+        
-            
+        self._network_type_callback = network_type_callback
-            self._network_type_callback = network_type_callback
+        self._on_poor_quality_callback = on_poor_quality_callback
-            self._on_poor_quality_callback = on_poor_quality_callback
+        self._wifi_quality_stop_event.clear()
-            self._wifi_quality_stop_event.clear()
+        self._wifi_quality_monitor_thread = threading.Thread(
-            self._wifi_quality_monitor_thread = threading.Thread(
+            target=self._quality_monitor_loop,
-                target=self._quality_monitor_loop,
+            daemon=True,
-                daemon=True,
+            name="wifi_quality_monitor"
-                name="wifi_quality_monitor"
+        )
-            )
+        self._wifi_quality_monitor_thread.start()
-            self._wifi_quality_monitor_thread.start()
+        self.logger.info("[WiFi Monitor] 已启动后台监测线程")
-            self.logger.info("[WiFi Monitor] 已启动后台监测线程")
     
     def stop_quality_monitor(self):
         """停止 WiFi 质量监测线程"""
-        with self._wifi_quality_lock:
+        if self._wifi_quality_monitor_thread is None:
-            t = self._wifi_quality_monitor_thread
+            return
-            if t is None:
-                return
-            if not t.is_alive():
-                self._wifi_quality_monitor_thread = None
-                return
         
         self._wifi_quality_stop_event.set()
         try:
-            t.join(timeout=2.0)
+            self._wifi_quality_monitor_thread.join(timeout=2.0)
         except Exception as e:
             self.logger.error(f"[WiFi Monitor] 停止线程失败：{e}")
-        
+        finally:
-        with self._wifi_quality_lock:
+            self._wifi_quality_monitor_thread = None
-            if t is self._wifi_quality_monitor_thread:
+            self.logger.info("[WiFi Monitor] 已停止后台监测线程")
-                if t.is_alive():
-                    self.logger.warning("[WiFi Monitor] 线程未在超时内退出，保留引用防止重复创建")
-                else:
-                    self._wifi_quality_monitor_thread = None
-                    self.logger.info("[WiFi Monitor] 已停止后台监测线程")
     
     def _quality_monitor_loop(self):
         """

yolo_te.py

@@ -0,0 +1,267 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""Standalone live camera + single YOLO runner.
+
+不复用项目内的 `camera_manager` / `target_roi_yolo` / `config` / `logger_manager`。
+
+功能：
+- 独立初始化摄像头
+- 实时读取帧
+- 独立加载单个 YOLO 模型并推理
+- 画出检测框、ROI、FPS
+
+适用场景：
+- 单独验证一个模型是否能跑
+- 验证实时帧率
+- 验证 ROI 是否裁对
+- 不进入主业务射箭流程
+"""
+
+from __future__ import annotations
+
+import os
+import time
+from dataclasses import dataclass
+
+
+@dataclass
+class RunnerConfig:
+    camera_width: int = 640
+    camera_height: int = 480
+    model_path: str = "/root/model_278702.mud"
+    conf_th: float = 0.7
+    retry_conf_th: float = 0.5
+    class_ids: tuple = (0,)
+    merge_mode: str = "union"
+    coord_mode: str = "native"
+    roi_margin_frac: float = 0.11
+    min_box_side_px: int = 8
+
+
+def log(msg: str):
+    print(msg)
+
+
+class DummyLogger:
+    def info(self, msg):
+        log(msg)
+
+    def warning(self, msg):
+        log(msg)
+
+    def error(self, msg):
+        log(msg)
+
+
+class StandaloneYOLORunner:
+    def __init__(self, cfg: RunnerConfig):
+        self.cfg = cfg
+        self.logger = DummyLogger()
+        self._last_fps_t = time.perf_counter()
+        self._frames = 0
+        self._fps = 0.0
+        self._camera = None
+        self._det = None
+
+    def _import_maix(self):
+        try:
+            from maix import camera, image, nn
+            return camera, image, nn
+        except Exception as e:
+            raise RuntimeError(f"maix import failed: {e}")
+
+    def _init_camera(self):
+        camera, _, _ = self._import_maix()
+        if self._camera is not None:
+            return self._camera
+        try:
+            self._camera = camera.Camera(
+                width=self.cfg.camera_width,
+                height=self.cfg.camera_height,
+                format=camera.RGB888,
+            )
+        except Exception:
+            self._camera = camera.Camera(width=self.cfg.camera_width, height=self.cfg.camera_height)
+        return self._camera
+
+    def _load_detector(self, model_path: str):
+        _, _, nn = self._import_maix()
+        if not model_path or not os.path.isfile(model_path):
+            return None
+        return nn.YOLOv5(model=model_path, dual_buff=False)
+
+    @staticmethod
+    def _get_class_id(obj):
+        for key in ("class_id", "cls", "label", "category", "cat_id", "id"):
+            if hasattr(obj, key):
+                v = getattr(obj, key)
+                if v is None:
+                    continue
+                try:
+                    return int(float(v))
+                except Exception:
+                    pass
+        return None
+
+    @staticmethod
+    def _normalize_boxes(raw):
+        out = []
+        for o in raw or []:
+            if isinstance(o, (list, tuple)) and len(o) >= 6:
+                class Box:
+                    pass
+                b = Box()
+                b.x, b.y, b.w, b.h, b.score, b.class_id = map(float, o[:6])
+                out.append(b)
+            else:
+                out.append(o)
+        return out
+
+    def _det_to_xyxy(self, det, obj):
+        x = float(getattr(obj, "x", 0.0))
+        y = float(getattr(obj, "y", 0.0))
+        w = float(getattr(obj, "w", 0.0))
+        h = float(getattr(obj, "h", 0.0))
+        return x, y, x + w, y + h
+
+    def _run_detector(self, det, img, conf_th, class_ids):
+        if det is None:
+            return []
+        raw = det.detect(img, conf_th=conf_th)
+        objs = self._normalize_boxes(raw if raw is not None else [])
+        out = []
+        for o in objs:
+            cid = self._get_class_id(o)
+            if cid is not None and cid not in class_ids:
+                continue
+            out.append(o)
+        return out
+
+    def _calc_fps(self):
+        self._frames += 1
+        now = time.perf_counter()
+        dt = now - self._last_fps_t
+        if dt >= 1.0:
+            self._fps = self._frames / dt
+            self._frames = 0
+            self._last_fps_t = now
+        return self._fps
+
+    def _draw_text(self, img, lines):
+        try:
+            import cv2
+            y = 24
+            for line in lines:
+                cv2.putText(img, line, (10, y), cv2.FONT_HERSHEY_SIMPLEX, 0.55, (0, 255, 0), 1, cv2.LINE_AA)
+                y += 20
+        except Exception:
+            pass
+
+    def _clip_roi(self, x0, y0, x1, y1, w, h):
+        x0 = max(0, min(int(x0), w - 1))
+        y0 = max(0, min(int(y0), h - 1))
+        x1 = max(x0 + 1, min(int(x1), w))
+        y1 = max(y0 + 1, min(int(y1), h))
+        return x0, y0, x1, y1
+
+    def _merge_boxes(self, boxes):
+        if not boxes:
+            return None
+        x0 = min(b[0] for b in boxes)
+        y0 = min(b[1] for b in boxes)
+        x1 = max(b[2] for b in boxes)
+        y1 = max(b[3] for b in boxes)
+        return x0, y0, x1, y1
+
+    def _run_single_yolo(self, frame, img_cv):
+        h, w = int(img_cv.shape[0]), int(img_cv.shape[1])
+        if self._det is None:
+            self._det = self._load_detector(self.cfg.model_path)
+        det = self._det
+        if det is None:
+            return []
+
+        boxes = self._run_detector(det, frame, self.cfg.conf_th, self.cfg.class_ids)
+        if not boxes and self.cfg.retry_conf_th < self.cfg.conf_th:
+            boxes = self._run_detector(det, frame, self.cfg.retry_conf_th, self.cfg.class_ids)
+
+        xyxy = []
+        for obj in boxes:
+            x0, y0, x1, y1 = self._det_to_xyxy(det, obj)
+            if (x1 - x0) < self.cfg.min_box_side_px or (y1 - y0) < self.cfg.min_box_side_px:
+                continue
+            if self.cfg.coord_mode == "native":
+                x0, y0, x1, y1 = self._clip_roi(x0, y0, x1, y1, w, h)
+            xyxy.append((x0, y0, x1, y1))
+        return xyxy
+
+    def run(self):
+        _, image, _ = self._import_maix()
+        cam = self._init_camera()
+        log("[YOLOTE] standalone runner started")
+
+        while True:
+            try:
+                frame = cam.read()
+            except Exception as e:
+                log(f"[YOLOTE] camera read failed: {e}")
+                time.sleep(0.02)
+                continue
+
+            if frame is None:
+                time.sleep(0.01)
+                continue
+
+            try:
+                img_cv = image.image2cv(frame, False, False)
+            except Exception as e:
+                log(f"[YOLOTE] image2cv failed: {e}")
+                time.sleep(0.01)
+                continue
+
+            import cv2
+            t0 = time.perf_counter()
+            boxes = self._run_single_yolo(frame, img_cv)
+            t1 = time.perf_counter()
+
+            for i, (bx0, by0, bx1, by1) in enumerate(boxes):
+                cv2.rectangle(img_cv, (int(bx0), int(by0)), (int(bx1) - 1, int(by1) - 1), (0, 255, 0), 2)
+                cv2.putText(img_cv, f"B{i}", (int(bx0), max(0, int(by0) - 4)), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 1, cv2.LINE_AA)
+
+            fps = self._calc_fps()
+            self._draw_text(
+                img_cv,
+                [
+                    f"FPS: {fps:.1f}",
+                    f"YOLO: {(t1 - t0)*1000.0:.1f} ms",
+                    f"Boxes: {len(boxes)}",
+                    "Ctrl+C to exit",
+                ],
+            )
+
+            try:
+                frame_out = image.cv2image(img_cv, False, False)
+                if hasattr(cam, "show"):
+                    cam.show(frame_out)
+                else:
+                    try:
+                        frame_out.show()
+                    except Exception:
+                        pass
+            except Exception as e:
+                log(f"[YOLOTE] show failed: {e}")
+
+            time.sleep(0.001)
+
+
+def main():
+    cfg = RunnerConfig()
+    runner = StandaloneYOLORunner(cfg)
+    try:
+        runner.run()
+    except KeyboardInterrupt:
+        log("[YOLOTE] interrupted")
+
+
+if __name__ == "__main__":
+    main()