feat: 2.17.18

.vscode/settings.json

@@ -0,0 +1,3 @@
+{
+    "cmake.sourceDirectory": "E:/code/code/code/new/new/new/new/new/nw/2.17.0/archery/cpp_ext"
+}

app.yaml

@@ -1,6 +1,6 @@
 id: t11
 name: t11
-version: 2.16.4
+version: 2.17.18
 author: t11
 icon: ''
 desc: t11
@@ -18,6 +18,8 @@ files:
   - laser_manager.py
   - logger_manager.py
   - main.py
+  - model_317828.cvimodel
+  - model_317828.mud
   - network.py
   - ota_curl.sh
   - ota_manager.py

config.py

@@ -234,10 +234,10 @@ TRIANGLE_BLACKHAT_KERNEL_FRAC = 0.018  # 核大小 ≈ min(h,w)*frac，取奇数
 # ── YOLO(NPU) 靶环 ROI → 裁剪后再跑三角形（减小 CPU 处理面积）──────────────────
 # 日志里 net_in=W×H 来自 .mud 模型（det.input_width/height），不是这里配置的。
 TRIANGLE_YOLO_ROI_ENABLE = True
-TRIANGLE_YOLO_MODEL_PATH = APP_DIR + "/model_270139.mud"
+TRIANGLE_YOLO_MODEL_PATH = APP_DIR + "/model_317211.mud"
 # 参与 ROI 的类别：多类时只填「整靶/靶环」的 id；不要填角标类，否则 union 仍可对，但 largest 会偏小。
 TRIANGLE_YOLO_RING_CLASS_IDS = (0,)
-TRIANGLE_YOLO_CONF_TH = 0.7
+TRIANGLE_YOLO_CONF_TH = 0.9
 TRIANGLE_YOLO_IOU_TH = 0.45
 # YOLO 首次/临界帧可能在高阈值下 0 框；启用后仅在 0 候选时用较低阈值重试一次。
 # 后续仍会经过 min_box_side、ROI aspect、三角形几何校验，避免直接放大假阳性。
@@ -262,6 +262,16 @@ TRIANGLE_SAMPLE_PATCH_HALF_PX = 2
 # 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
 
+# YOLO target size classification: class 0=20cm, class 1=40cm.
+TARGET_CLASS_YOLO_ENABLE = True
+TARGET_CLASS_YOLO_MODEL_PATH = APP_DIR + "/model_317828.mud"
+TARGET_CLASS_YOLO_LABELS = (20, 40)
+TARGET_CLASS_YOLO_CONF_TH = 0.66
+TARGET_CLASS_YOLO_IOU_TH = 0.45
+TARGET_CLASS_YOLO_RETRY_ON_EMPTY = False
+TARGET_CLASS_YOLO_RETRY_CONF_TH = 0.25
+TARGET_CLASS_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）：
 # "yolo" — 调 Stage2 黑三角模型得子框，再子框内传统提取（需 TRIANGLE_BLACK_YOLO_ENABLE=True）。
@@ -316,10 +326,13 @@ 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_IMAGE_ON_FAILURE = True  # 检测失败时是否强制保存图像（供调试测试用）
 PHOTO_DIR = "/root/phot"    # 照片存储目录
 MAX_IMAGES = 1000
+SAVE_RAW_IMAGE_ENABLED = False  # 额外保存完整原始帧（不画框、不画点、不裁剪）
+RAW_IMAGE_DIR = PHOTO_DIR + "/raw"
+RAW_IMAGE_MAX_IMAGES = MAX_IMAGES
 # 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
 

main.py

@@ -132,6 +132,7 @@ def cmd_str():
     sync_system_time_from_4g()
 
     # 2.1 WiFi 热点配网兜底：仅当 STA 与 4G 均不可用时起 AP + HTTP；提交后删 /boot/wifi.ap、建 wifi.sta 并 reboot
+    _ota_pending_path = f"{config.APP_DIR}/ota_pending.json"
     try:
         from wifi_config_httpd import maybe_start_wifi_ap_fallback
 
@@ -162,15 +163,21 @@ def cmd_str():
             and _loc_black == "yolo"
             and bool(getattr(config, "TRIANGLE_BLACK_YOLO_PRELOAD_ON_BOOT", True))
         )
-        _preload_yolo = _preload_yolo or _need_black_preload
+        _need_target_preload = (
-        if _preload_yolo:
+            bool(getattr(config, "TARGET_CLASS_YOLO_ENABLE", False))
+            and bool(getattr(config, "TARGET_CLASS_YOLO_PRELOAD_ON_BOOT", True))
+        )
+        _preload_yolo = _preload_yolo or _need_black_preload or _need_target_preload
+        if _preload_yolo and not os.path.exists(_ota_pending_path):
             preload_yolo_detector(logger)
+        elif _preload_yolo and logger:
+            logger.warning("[YOLO] ota_pending.json found; skip model preload until rollback check")
     except Exception as e:
         if logger:
             logger.warning(f"[YOLO-ROI] 启动预加载异常（不影响后续射箭）: {e}")
 
     # 3. 启动时检查：是否需要恢复备份
-    pending_path = f"{config.APP_DIR}/ota_pending.json"
+    pending_path = _ota_pending_path
     if os.path.exists(pending_path):
         try:
             with open(pending_path, 'r', encoding='utf-8') as f:
@@ -246,7 +253,11 @@ def cmd_str():
     network_manager.read_device_id()
 
     # 5. 创建照片存储目录（如果启用图像保存或检测失败时强制保存）
-    if config.SAVE_IMAGE_ENABLED or getattr(config, "SAVE_IMAGE_ON_FAILURE", False):
+    if (
+        config.SAVE_IMAGE_ENABLED
+        or getattr(config, "SAVE_IMAGE_ON_FAILURE", False)
+        or getattr(config, "SAVE_RAW_IMAGE_ENABLED", False)
+    ):
         photo_dir = config.PHOTO_DIR
         if photo_dir not in os.listdir("/root"):
             try:
@@ -278,12 +289,13 @@ def cmd_str():
         logger.info("系统准备完成...")
 
     last_adc_trigger = 0
+    trigger_adc_val = 0  # 触发时的气压值，气压需降回此值以下才能再次触发
     # 读取一次ADC初始值，防止开机时传感器已有压力导致误触发
+    enable_check = True
     try:
         last_adc_val = hardware_manager.adc_obj.read()
     except Exception:
         last_adc_val = 0
-    peak_adc_val = 0  # 当前周期内的压力峰值
     # 气压采样：减少日志频率（每 N 个点输出一条），避免 logger.debug 拖慢采样
     PRESSURE_BATCH_SIZE = 100
 
@@ -373,22 +385,16 @@ def cmd_str():
                 pressure_max = adc_val
             if len(pressure_buf) >= PRESSURE_BATCH_SIZE:
                 _flush_pressure_buf("batch")
-            # 峰值检测：压力从峰值下降时触发，确保捕获到最大冲击时刻
+            # 突变增量检测：压力增量大于400时触发
-            if adc_val > peak_adc_val:
+            # 触发后需等气压降到触发值以下才重新检测增量
-                peak_adc_val = adc_val  # 更新峰值
+            if adc_val < trigger_adc_val :
-            if (peak_adc_val >= config.ADC_TRIGGER_THRESHOLD
+                enable_check = True
-                    and adc_val < peak_adc_val
+            if (adc_val - last_adc_val) > 200 and enable_check:
-                    and last_adc_val >= peak_adc_val):
-                # 封顶后下降沿触发：peak是最大值，当前值开始下降，且上次值还在peak位置
                 hardware_manager.start_idle_timer()  # 重新计时
-                diff_ms = current_time - last_adc_trigger
-                if diff_ms < 3000:
-                    peak_adc_val = 0  # 去抖期间重置峰值
-                    time.sleep_ms(5)
-                    continue
                 last_adc_trigger = current_time
-                peak_adc_val = 0  # 触发后重置峰值
+                trigger_adc_val = adc_val  # 记录触发时的气压值
-                # 触发前先把缓存刷出来，避免波形被长耗时处理截断
+                last_adc_val = adc_val  # 更新基准值，防止连续增量误触发
+                enable_check = False
                 _flush_pressure_buf("before_trigger")
 
                 try:
@@ -407,7 +413,7 @@ def cmd_str():
                         camera_manager.show(camera_manager.read_frame())
                     except Exception as e:
                         pass
-                time.sleep_ms(5)
+                time.sleep_ms(1)
             last_adc_val = adc_val
 
         except Exception as e:

model_317189.mud

@@ -1,7 +1,7 @@
 
 [basic]
 type = cvimodel
-model = model_270139.cvimodel
+model = model_317189.cvimodel
 
 [extra]
 model_type = yolov5
@@ -9,5 +9,5 @@ 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 = 黑三角和圆环
+labels = circle, triangle
 

model_317211.mud

@@ -0,0 +1,13 @@
+
+[basic]
+type = cvimodel
+model = model_317211.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 = circle, triangle
+

model_317423.mud

@@ -0,0 +1,13 @@
+
+[basic]
+type = cvimodel
+model = model_317423.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 = 20, 10, 40
+

model_317704.mud

@@ -0,0 +1,13 @@
+
+[basic]
+type = cvimodel
+model = model_317704.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 = 40, circle, triangle
+

model_317828.mud

@@ -1,7 +1,7 @@
 
 [basic]
 type = cvimodel
-model = model_270820.cvimodel
+model = model_317828.cvimodel
 
 [extra]
 model_type = yolov5
@@ -9,5 +9,5 @@ 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
+labels = 20, 40
 

shoot_manager.py

@@ -8,7 +8,12 @@ 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
@@ -54,6 +59,7 @@ def analyze_shot(frame, laser_point=None):
     """
     logger = logger_manager.logger
     from datetime import datetime
+    yellow_algorithm_ms = 0.0
 
     # ── Step 1: 确定激光点 ────────────────────────────────────────────────────
     laser_point_method = None
@@ -69,7 +75,11 @@ def analyze_shot(frame, laser_point=None):
             logger.info(f"[算法] 使用校准值: {laser_manager.laser_point}")
     else:
         # 动态模式：先做一次无激光点检测以估算距离，再推算激光点
+        _t_yellow = time_std.perf_counter()
+        try:
             _, _, _, _, best_radius1_temp, _ = detect_circle_v3(frame, None)
+        finally:
+            yellow_algorithm_ms += (time_std.perf_counter() - _t_yellow) * 1000.0
         distance_m_first = estimate_distance(best_radius1_temp) if best_radius1_temp else None
         if distance_m_first and distance_m_first > 0:
             laser_point = laser_manager.calculate_laser_point_from_distance(distance_m_first)
@@ -114,6 +124,7 @@ 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",
+            "yellow_algorithm_ms": float(yellow_algorithm_ms),
         }
         if yolo_roi_xyxy is not None:
             out["yolo_roi_xyxy"] = yolo_roi_xyxy
@@ -121,9 +132,12 @@ def analyze_shot(frame, laser_point=None):
 
     if not use_tri:
         # 三角形未配置，直接跑圆形检测
-        return _build_circle_result(
+        _t_yellow = time_std.perf_counter()
-            detect_circle_v3(frame, laser_point, img_cv=img_cv)
+        try:
-        )
+            cdata = detect_circle_v3(frame, laser_point, img_cv=img_cv)
+        finally:
+            yellow_algorithm_ms += (time_std.perf_counter() - _t_yellow) * 1000.0
+        return _build_circle_result(cdata)
 
     # ── Step 4: 先独占跑三角形，超时或失败后再跑圆形（不与圆心并行，避免抢 CPU）──
     roi_xyxy = None
@@ -276,6 +290,7 @@ def analyze_shot(frame, laser_point=None):
             "laser_point": laser_point, "laser_point_method": laser_point_method,
             "offset_method": tri.get("offset_method") or "triangle_homography",
             "distance_method": tri.get("distance_method") or "pnp_triangle",
+            "yellow_algorithm_ms": float(yellow_algorithm_ms),
             "tri_markers": tri.get("markers", []),
             "tri_markers_completed": tri.get("markers_completed", []),
             "tri_homography": tri.get("homography"),
@@ -295,8 +310,12 @@ def analyze_shot(frame, laser_point=None):
         logger.warning(f"[TRI] 超时 {tri_timeout_s:.2f}s 仍未结束，启动圆心算法（三角形仍在后台）")
 
     # 三角形超时或失败 → 跑圆心；圆心跑完后再检查三角形是否已结束
+    try:
+        _t_yellow = time_std.perf_counter()
         try:
             cdata = detect_circle_v3(frame, laser_point, img_cv=img_cv)
+        finally:
+            yellow_algorithm_ms += (time_std.perf_counter() - _t_yellow) * 1000.0
     except Exception as e:
         logger.error(f"[CIRCLE] 圆形检测异常: {e}")
         cdata = (frame, None, None, None, None, None)
@@ -322,9 +341,31 @@ def process_shot(adc_val):
     try:
         frame = camera_manager.read_frame()
 
+        # Copy the untouched frame before any detection or drawing.
+        from shot_id_generator import shot_id_generator
+        shot_id = shot_id_generator.generate_id()
+        enqueue_save_raw_shot(frame, shot_id)
+
         # 网络事件移到拍照之后，避免阻塞拍照
         network_manager.safe_enqueue({"shoot_event": "start"}, msg_type=2, high=True)
 
+        # Classify only the current shot frame; never reuse a previous result.
+        target_class_result = None
+        yolo_target_ms = 0.0
+        try:
+            from target_roi_yolo import try_get_target_class_from_yolo
+
+            _t_yolo_target = time_std.perf_counter()
+            try:
+                target_class_result = try_get_target_class_from_yolo(frame, logger=logger)
+            finally:
+                yolo_target_ms = (time_std.perf_counter() - _t_yolo_target) * 1000.0
+            if logger:
+                logger.info(f"[YOLO-TARGET] 当前箭业务结果: {target_class_result}")
+        except Exception as exc:
+            if logger:
+                logger.warning(f"[YOLO-TARGET] 当前箭分类失败，按未知处理: {exc}")
+
         # 调用算法分析
         analysis_result = analyze_shot(frame)
 
@@ -348,6 +389,7 @@ def process_shot(adc_val):
         laser_point_method = analysis_result["laser_point_method"]
         offset_method = analysis_result.get("offset_method", "yellow_circle")
         distance_method = analysis_result.get("distance_method", "yellow_radius")
+        yellow_algorithm_ms = float(analysis_result.get("yellow_algorithm_ms", 0.0) or 0.0)
         tri_markers = analysis_result.get("tri_markers", [])
         tri_markers_completed = analysis_result.get("tri_markers_completed", [])
         tri_homography = analysis_result.get("tri_homography")
@@ -368,10 +410,6 @@ 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}")
 
@@ -384,11 +422,27 @@ def process_shot(adc_val):
         srv_y = round(float(dy), 4) if dy is not None else 200.0
 
         # 构造上报数据
+        target_label = (
+            target_class_result.get("label")
+            if isinstance(target_class_result, dict)
+            else None
+        )
+        target_confidence = (
+            target_class_result.get("confidence")
+            if isinstance(target_class_result, dict)
+            else None
+        )
         inner_data = {
             "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）
+            "target_class": target_label,
+            "target_class_confidence": (
+                round(float(target_confidence), 2)
+                if target_confidence is not None
+                else None
+            ),
             "d": round((distance_m or 0.0) * 100),
             "d_laser": round((laser_distance_m or 0.0) * 100),
             "d_laser_quality": laser_signal_quality,
@@ -399,6 +453,8 @@ def process_shot(adc_val):
             "target_y": float(y),
             "offset_method": offset_method,
             "distance_method": distance_method,
+            "yellow_algorithm_ms": round(yellow_algorithm_ms, 2),
+            "yolo_target_ms": round(float(yolo_target_ms), 2),
         }
 
         if ellipse_params:
@@ -415,7 +471,19 @@ def process_shot(adc_val):
             inner_data["ellipse_center_x"] = None
             inner_data["ellipse_center_y"] = None
 
+        # 记录这组 inner_data 即将进入上报队列的本地时间，精确到毫秒。
+        upload_time_ms = int(time_std.time() * 1000)
+        upload_time_sec, upload_time_millis = divmod(upload_time_ms, 1000)
+        inner_data["upload_time"] = (
+            time_std.strftime("%Y-%m-%d %H:%M:%S", time_std.localtime(upload_time_sec))
+            + f".{upload_time_millis:03d}"
+        )
         report_data = {"cmd": 1, "data": inner_data}
+        if logger:
+            logger.info(
+                f"[REPORT-TARGET] enqueue shot_id={shot_id}, "
+                f"target_class={target_label}, confidence={target_confidence}"
+            )
         network_manager.safe_enqueue(report_data, msg_type=2, high=True)
 
         # 数据上报后再画标注，不干扰检测阶段的原始画面

target_roi_yolo.py

@@ -89,6 +89,29 @@ def _stage2_roi_crop_save_worker(
 _detector_by_path = {}
 
 
+def _resolve_model_path(model_path: str):
+    """Resolve a model in either the installed app or MaixVision run directory."""
+    model_path = (model_path or "").strip()
+    if model_path and os.path.isfile(model_path):
+        return model_path
+    if not model_path:
+        return ""
+    name = os.path.basename(model_path)
+    module_dir = os.path.dirname(os.path.abspath(__file__))
+    candidates = (
+        os.path.join(module_dir, name),
+        os.path.join(module_dir, "test", name),
+        os.path.join("/tmp/maixpy_run", name),
+        os.path.join("/tmp/maixpy_run", "test", name),
+        os.path.join(os.getcwd(), name),
+        os.path.join(os.getcwd(), "test", name),
+    )
+    for candidate in candidates:
+        if os.path.isfile(candidate):
+            return candidate
+    return model_path
+
+
 def reset_yolo_detector_cache():
     """切换模型路径时可调用（通常不必）。"""
     global _detector_by_path
@@ -103,10 +126,19 @@ def _get_detector(model_path: str):
         return _detector_by_path[model_path]
     try:
         from maix import nn
-    except ImportError:
+    except Exception:
         return None
-    _detector_by_path[model_path] = nn.YOLOv5(model=model_path, dual_buff=False)
+    # YOLO is an optional capability. A broken/incompatible model must not
-    return _detector_by_path[model_path]
+    # abort boot (especially before the OTA rollback check).
+    try:
+        detector = nn.YOLOv5(model=model_path, dual_buff=False)
+    except Exception:
+        # Cache the failure to avoid retrying a broken native load every frame.
+        # reset_yolo_detector_cache() clears this after a model replacement.
+        _detector_by_path[model_path] = None
+        return None
+    _detector_by_path[model_path] = detector
+    return detector
 
 
 def preload_yolo_detector(logger=None):
@@ -175,6 +207,23 @@ def preload_yolo_detector(logger=None):
                 % (_loc_black,)
             )
 
+    if bool(getattr(cfg, "TARGET_CLASS_YOLO_ENABLE", False)) and bool(
+        getattr(cfg, "TARGET_CLASS_YOLO_PRELOAD_ON_BOOT", True)
+    ):
+        class_model_path = _resolve_model_path(
+            getattr(cfg, "TARGET_CLASS_YOLO_MODEL_PATH", "") or ""
+        )
+        class_detector = _get_detector(class_model_path)
+        if class_detector is None:
+            if logger:
+                logger.warning(
+                    f"[YOLO-TARGET] 预加载失败：无法加载模型 {class_model_path}"
+                )
+        else:
+            ok = True
+            if logger:
+                logger.info(f"[YOLO-TARGET] 靶规格模型已预加载: {class_model_path}")
+
     return ok
 
 
@@ -206,8 +255,10 @@ def _det_obj_class_id(o):
             if v is None:
                 continue
             try:
+                if callable(v):
+                    v = v()
                 return int(float(v))
-            except (TypeError, ValueError):
+            except (TypeError, ValueError, AttributeError):
                 continue
     return None
 
@@ -242,6 +293,106 @@ def _normalize_objs(objs):
     return out
 
 
+def _det_obj_score(o):
+    """Return confidence across supported Maix YOLO result formats."""
+    for key in ("score", "confidence", "conf", "prob"):
+        if hasattr(o, key):
+            try:
+                value = getattr(o, key)
+                if callable(value):
+                    value = value()
+                value = float(value)
+                if value == value:
+                    return value
+            except (TypeError, ValueError, AttributeError):
+                pass
+    return 0.0
+
+
+def try_get_target_class_from_yolo(maix_frame, logger=None):
+    """Classify the current target as 20cm or 40cm; return None if unknown."""
+    try:
+        import config as cfg
+    except Exception:
+        return None
+
+    if not bool(getattr(cfg, "TARGET_CLASS_YOLO_ENABLE", False)):
+        return None
+    model_path = _resolve_model_path(
+        getattr(cfg, "TARGET_CLASS_YOLO_MODEL_PATH", "") or ""
+    )
+    if not os.path.isfile(model_path):
+        if logger:
+            logger.warning(f"[YOLO-TARGET] 模型文件不存在: {model_path}")
+        return None
+    detector = _get_detector(model_path)
+    if detector is None:
+        if logger:
+            logger.warning("[YOLO-TARGET] 无法加载 nn.YOLOv5")
+        return None
+
+    conf_th = float(getattr(cfg, "TARGET_CLASS_YOLO_CONF_TH", 0.5))
+    iou_th = float(getattr(cfg, "TARGET_CLASS_YOLO_IOU_TH", 0.45))
+    labels = getattr(cfg, "TARGET_CLASS_YOLO_LABELS", (20, 40))
+    if isinstance(labels, str):
+        labels = tuple(x.strip() for x in labels.split(",") if x.strip())
+    labels = tuple(labels)
+
+    def _detect(threshold):
+        try:
+            raw = detector.detect(maix_frame, conf_th=threshold, iou_th=iou_th)
+        except Exception as exc:
+            if logger:
+                logger.warning(f"[YOLO-TARGET] detect 异常: {exc}")
+            return []
+        return _normalize_objs(raw if raw is not None else [])
+
+    def _candidates(objs):
+        found = []
+        for obj in objs:
+            class_id = _det_obj_class_id(obj)
+            if class_id is None or class_id < 0 or class_id >= len(labels):
+                continue
+            try:
+                label = int(float(labels[class_id]))
+            except (TypeError, ValueError):
+                continue
+            if label in (20, 40):
+                found.append((label, class_id, _det_obj_score(obj)))
+        return found
+
+    objects = _detect(conf_th)
+    candidates = _candidates(objects)
+    if logger and objects:
+        logger.info(
+            "[YOLO-TARGET] 原始框=%d, 解析类别=%s"
+            % (
+                len(objects),
+                [(_det_obj_class_id(o), _det_obj_score(o)) for o in objects[:8]],
+            )
+        )
+    if not candidates and bool(
+        getattr(cfg, "TARGET_CLASS_YOLO_RETRY_ON_EMPTY", False)
+    ):
+        retry_th = float(getattr(cfg, "TARGET_CLASS_YOLO_RETRY_CONF_TH", conf_th))
+        if 0 < retry_th < conf_th:
+            candidates = _candidates(_detect(retry_th))
+
+    if not candidates:
+        if logger:
+            logger.warning("[YOLO-TARGET] 当前帧未识别到 20/40，按未知处理")
+        return None
+
+    label, class_id, confidence = max(candidates, key=lambda item: item[2])
+    result = {"label": label, "class_id": class_id, "confidence": confidence}
+    if logger:
+        logger.info(
+            f"[YOLO-TARGET] 当前帧分类={label}, class_id={class_id}, "
+            f"conf={confidence:.3f}"
+        )
+    return result
+
+
 def _det_to_src_xyxy(o, coord_mode: str, src_w: int, src_h: int, net_w: int, net_h: int):
     """把单个检测框转为全图坐标系下的 xyxy（半开区间语义与后续 clip 一致）。"""
     x, y, w, h = float(o.x), float(o.y), float(o.w), float(o.h)

test/test_target_yolo_maixvision.py

@@ -0,0 +1,108 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""Run from MaixVision on PC to inspect the box's live 20/40 YOLO output."""
+
+import os
+
+from maix import app, camera, display, image, nn, time
+
+
+# This file is sent to /tmp/maixpy_run by MaixVision. Keep the model path
+# absolute so the script uses the model already installed on the box.
+MODEL_PATH = "/maixapp/apps/t11/model_317181.mud"
+CAMERA_WIDTH = 640
+CAMERA_HEIGHT = 480
+CONF_TH = 0.65
+IOU_TH = 0.45
+
+
+def _flatten_objects(raw):
+    if raw is None:
+        return []
+    if isinstance(raw, (list, tuple)):
+        result = []
+        for item in raw:
+            if isinstance(item, (list, tuple)):
+                result.extend(_flatten_objects(item))
+            else:
+                result.append(item)
+        return result
+    return [raw]
+
+
+def main():
+    if not os.path.isfile(MODEL_PATH):
+        raise FileNotFoundError("model not found on box: " + MODEL_PATH)
+
+    detector = nn.YOLOv5(model=MODEL_PATH, dual_buff=False)
+    cam = camera.Camera(CAMERA_WIDTH, CAMERA_HEIGHT)
+    disp = display.Display()
+
+    labels = tuple(str(label) for label in detector.labels)
+    print("[YOLO] model:", MODEL_PATH)
+    print("[YOLO] labels:", labels)
+    print("[YOLO] conf=%.2f iou=%.2f" % (CONF_TH, IOU_TH))
+
+    fps = 0.0
+    frame_count = 0
+    last_log_ms = time.ticks_ms()
+
+    while not app.need_exit():
+        loop_start_ms = time.ticks_ms()
+        img = cam.read()
+
+        detect_start_ms = time.ticks_ms()
+        raw = detector.detect(img, conf_th=CONF_TH, iou_th=IOU_TH)
+        detect_ms = max(0, time.ticks_diff(time.ticks_ms(), detect_start_ms))
+        objects = _flatten_objects(raw)
+
+        candidates = []
+        for obj in objects:
+            class_id = int(obj.class_id)
+            score = float(obj.score)
+            label = labels[class_id] if 0 <= class_id < len(labels) else "unknown"
+            color = image.COLOR_GREEN if label in ("20", "40") else image.COLOR_RED
+
+            img.draw_rect(obj.x, obj.y, obj.w, obj.h, color=color)
+            img.draw_string(
+                obj.x,
+                max(0, obj.y - 16),
+                "%scm %.2f" % (label, score),
+                color=color,
+            )
+            if label in ("20", "40"):
+                candidates.append((score, label))
+
+        loop_ms = max(1, time.ticks_diff(time.ticks_ms(), loop_start_ms))
+        instant_fps = 1000.0 / float(loop_ms)
+        fps = instant_fps if frame_count == 0 else fps * 0.9 + instant_fps * 0.1
+
+        if candidates:
+            best_score, best_label = max(candidates, key=lambda item: item[0])
+            status = "TARGET %scm %.2f" % (best_label, best_score)
+            status_color = image.COLOR_GREEN
+        else:
+            status = "TARGET UNKNOWN"
+            status_color = image.COLOR_RED
+
+        img.draw_string(5, 5, status, color=status_color)
+        img.draw_string(
+            5,
+            25,
+            "infer=%dms fps=%.1f boxes=%d" % (detect_ms, fps, len(objects)),
+            color=image.COLOR_YELLOW,
+        )
+        disp.show(img)
+
+        frame_count += 1
+        now_ms = time.ticks_ms()
+        if time.ticks_diff(now_ms, last_log_ms) >= 1000:
+            print(
+                "[YOLO] %s infer=%dms fps=%.1f boxes=%d"
+                % (status, detect_ms, fps, len(objects))
+            )
+            last_log_ms = now_ms
+
+
+if __name__ == "__main__":
+    main()

test_traditional_target.py

@@ -0,0 +1,184 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""Offline baseline for traditional target-paper detection.
+
+Dataset format: sibling .txt files use YOLO boxes and classes.txt maps ids
+(the supplied dataset uses 0=40, 1=20, 2=10).  This intentionally simple
+baseline uses grayscale segmentation and contour geometry; it is useful as a
+reference before adding more specialized black-triangle grouping.
+"""
+from __future__ import annotations
+
+import argparse
+import csv
+import glob
+import itertools
+import os
+
+import cv2
+import numpy as np
+
+
+def detect_white_papers(image: np.ndarray) -> list[tuple[int, int, int, int]]:
+    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
+    h, w = gray.shape[:2]
+    mask = cv2.inRange(gray, 120, 255)
+    mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, np.ones((9, 9), np.uint8))
+    mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, np.ones((5, 5), np.uint8))
+    contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+    out = []
+    for contour in contours:
+        x, y, bw, bh = cv2.boundingRect(contour)
+        area = float(bw * bh)
+        if area < 0.05 * w * h or min(bw, bh) < 80:
+            continue
+        fill = cv2.contourArea(contour) / max(area, 1.0)
+        aspect = bw / max(float(bh), 1.0)
+        if fill >= 0.45 and 0.4 <= aspect <= 2.5:
+            out.append((x, y, x + bw, y + bh))
+    return out
+
+
+def detect_black_triangle_papers(image: np.ndarray):
+    """Infer paper boxes from the four small black corner marks."""
+    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
+    mask = cv2.inRange(gray, 0, 100)
+    mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, np.ones((2, 2), np.uint8))
+    contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+    points = []
+    for contour in contours:
+        x, y, bw, bh = cv2.boundingRect(contour)
+        area = cv2.contourArea(contour)
+        vertices = cv2.approxPolyDP(contour, 0.08 * cv2.arcLength(contour, True), True)
+        if 60 <= area <= 400 and 8 <= bw <= 24 and 8 <= bh <= 24:
+            if 3 <= len(vertices) <= 5 and 0.5 <= bw / max(bh, 1) <= 2.0:
+                points.append((x + bw / 2.0, y + bh / 2.0))
+    candidates = []
+    for group in itertools.combinations(points, 4):
+        xs = sorted(p[0] for p in group)
+        ys = sorted(p[1] for p in group)
+        span_x, span_y = xs[-1] - xs[0], ys[-1] - ys[0]
+        if span_x < 50 or span_y < 50 or not 0.45 < span_x / span_y < 1.5:
+            continue
+        corners = ((xs[0], ys[0]), (xs[-1], ys[0]),
+                   (xs[0], ys[-1]), (xs[-1], ys[-1]))
+        error = max(min(np.hypot(p[0] - c[0], p[1] - c[1]) for c in corners)
+                    for p in group) / max(span_x, span_y)
+        if error > 0.22:
+            continue
+        ex, ey = 0.12 * span_x, 0.12 * span_y
+        candidates.append((xs[0] - ex, ys[0] - ey,
+                           xs[-1] + ex, ys[-1] + ey, error))
+    # A colored target ring supplies an independent center check. Hough is
+    # deliberately low-cost here because it runs only on the already small
+    # candidate list's source frame.
+    hsv = cv2.cvtColor(image, cv2.COLOR_BGR2HSV)
+    color = cv2.inRange(hsv, (0, 70, 45), (179, 255, 255))
+    color = cv2.morphologyEx(color, cv2.MORPH_OPEN, np.ones((5, 5), np.uint8))
+    ring_centers = []
+    for contour in cv2.findContours(color, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)[0]:
+        area = cv2.contourArea(contour)
+        if area < 150:
+            continue
+        moments = cv2.moments(contour)
+        if moments["m00"]:
+            ring_centers.append((moments["m10"] / moments["m00"], moments["m01"] / moments["m00"]))
+    checked = []
+    for box in candidates:
+        if not ring_centers:
+            checked.append(box)
+            continue
+        x0, y0, x1, y1, err = box
+        inside = any(x0 - .15 * (x1 - x0) <= cx <= x1 + .15 * (x1 - x0)
+                     and y0 - .15 * (y1 - y0) <= cy <= y1 + .15 * (y1 - y0)
+                     for cx, cy in ring_centers)
+        if inside:
+            checked.append(box)
+    return sorted(checked, key=lambda x: x[-1])
+
+
+def iou(a, b):
+    x0, y0 = max(a[0], b[0]), max(a[1], b[1])
+    x1, y1 = min(a[2], b[2]), min(a[3], b[3])
+    inter = max(0, x1 - x0) * max(0, y1 - y0)
+    aa = max(0, a[2] - a[0]) * max(0, a[3] - a[1])
+    bb = max(0, b[2] - b[0]) * max(0, b[3] - b[1])
+    return inter / max(aa + bb - inter, 1)
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    ap.add_argument("dataset", help="directory containing jpg and YOLO txt files")
+    ap.add_argument("--iou", type=float, default=0.5)
+    ap.add_argument("--out", default="traditional_eval_results.csv",
+                    help="CSV output path; relative paths are next to the dataset")
+    ap.add_argument("--vis-dir", default="traditional_eval_images",
+                    help="directory for annotated result images; empty disables")
+    args = ap.parse_args()
+    stats = {0: [0, 0], 1: [0, 0]}
+    rows = []
+    # OpenCV on some Windows builds cannot decode non-ASCII filenames. Work
+    # relative to the dataset directory so the supplied Chinese path is safe.
+    dataset = os.path.abspath(args.dataset)
+    os.chdir(dataset)
+    # cwd is now the dataset, so a relative output avoids Windows console
+    # encoding issues with the Chinese parent path.
+    vis_dir = args.vis_dir if args.vis_dir else ""
+    if vis_dir:
+        os.makedirs(vis_dir, exist_ok=True)
+    files = glob.glob(os.path.join("**", "*.jpg"), recursive=True)
+    for image_path in files:
+        label_path = os.path.splitext(image_path)[0] + ".txt"
+        if not os.path.isfile(label_path):
+            continue
+        image = cv2.imread(image_path)
+        if image is None:
+            continue
+        h, w = image.shape[:2]
+        predictions = detect_black_triangle_papers(image)
+        vis = image.copy()
+        for p in predictions:
+            cv2.rectangle(vis, (int(p[0]), int(p[1])), (int(p[2]), int(p[3])), (0, 255, 255), 2)
+        for line in open(label_path, encoding="utf-8", errors="ignore"):
+            z = line.split()
+            if len(z) < 5 or int(float(z[0])) not in stats:
+                continue
+            cls, cx, cy, bw, bh = int(float(z[0])), *map(float, z[1:5])
+            truth = (int((cx - bw / 2) * w), int((cy - bh / 2) * h),
+                     int((cx + bw / 2) * w), int((cy + bh / 2) * h))
+            best = max((iou(truth, p) for p in predictions), default=0.0)
+            best_box = max(predictions, key=lambda p: iou(truth, p), default=())
+            stats[cls][0] += 1
+            stats[cls][1] += int(best >= args.iou)
+            rows.append({
+                "image": image_path,
+                "class_id": cls,
+                "truth_xyxy": ",".join(map(str, truth[:4])),
+                "pred_xyxy": ",".join(map(str, best_box[:4])) if best_box else "",
+                "iou": f"{best:.4f}",
+                "pass": int(best >= args.iou),
+            })
+            color = (0, 255, 0) if best >= args.iou else (0, 0, 255)
+            cv2.rectangle(vis, truth[:2], truth[2:4], color, 2)
+            cv2.putText(vis, f"GT {cls} IoU {best:.2f}",
+                        (truth[0], max(16, truth[1] - 4)),
+                        cv2.FONT_HERSHEY_SIMPLEX, 0.45, color, 1, cv2.LINE_AA)
+        if vis_dir:
+            name = os.path.splitext(os.path.basename(image_path))[0] + "_result.jpg"
+            cv2.imwrite(os.path.join(vis_dir, name), vis)
+    total = sum(v[0] for v in stats.values())
+    good = sum(v[1] for v in stats.values())
+    print(f"paper objects: {good}/{total} = {good / max(total, 1):.2%} (IoU >= {args.iou})")
+    for cls, (n, ok) in stats.items():
+        print(f"class {cls}: {ok}/{n} = {ok / max(n, 1):.2%}")
+    out_path = args.out if os.path.isabs(args.out) else os.path.join(dataset, args.out)
+    with open(out_path, "w", newline="", encoding="utf-8-sig") as fp:
+        writer = csv.DictWriter(fp, fieldnames=("image", "class_id", "truth_xyxy",
+                                                "pred_xyxy", "iou", "pass"))
+        writer.writeheader()
+        writer.writerows(rows)
+    print(f"details csv: {out_path}")
+
+
+if __name__ == "__main__":
+    main()

version.md

@@ -29,3 +29,12 @@
 # 2.15.16 修复wifi连接问题
 # 2.15.17 修复wifi连接问题
 # 2.15.18 wifi连接成功重新登录
+# 2.16.4  优化射箭延迟
+# 2.17.0  yolo标靶类别识别
+# 2.17.1  26-08-19 17：39 压力传感修改 增量方式 
+
+# 2.17.2 26-08-24 17：56 靶纸识别模型更替
+
+# 2.17.3 26-08-25 9：57 原图拍摄开关
+
+# 2.17.4 26-08-25 14：57 模型修改

version.py

@@ -4,6 +4,6 @@
 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 = '2.16.4'
+VERSION = '2.17.18'
 
 

vision.py

@@ -908,6 +908,11 @@ def _save_worker_loop():
             item = _save_queue.get()
             if item is None:
                 break
+            if isinstance(item, dict) and item.get("kind") == "raw":
+                _save_raw_image_impl(
+                    item["img_cv"], item["shot_id"], item["photo_dir"]
+                )
+            else:
                 _save_shot_image_impl(*item)
         except Exception as e:
             logger = logger_manager.logger
@@ -936,6 +941,56 @@ def start_save_shot_worker():
         logger.info("[VISION] 存图 worker 线程已启动")
 
 
+def _save_raw_image_impl(img_cv, shot_id, photo_dir):
+    """保存相机完整原始帧，不添加任何检测标注。"""
+    logger = logger_manager.logger
+    try:
+        os.makedirs(photo_dir, exist_ok=True)
+        filename = os.path.join(photo_dir, f"shot_{shot_id}_raw.jpg")
+        image.cv2image(img_cv, False, False).save(filename)
+        prune_old_images_in_dir(
+            photo_dir,
+            getattr(config, "RAW_IMAGE_MAX_IMAGES", config.MAX_IMAGES),
+            logger,
+            "[VISION-RAW]",
+        )
+        if logger:
+            logger.info(f"[VISION-RAW] 已保存纯原图: {filename}")
+        return filename
+    except Exception as e:
+        if logger:
+            logger.error(f"[VISION-RAW] 保存纯原图失败: {e}")
+        return None
+
+
+def enqueue_save_raw_shot(frame, shot_id, photo_dir=None):
+    """立即复制相机帧并异步保存，避免后续识别和绘图修改原图。"""
+    if not getattr(config, "SAVE_RAW_IMAGE_ENABLED", False):
+        return
+    if photo_dir is None:
+        photo_dir = getattr(
+            config, "RAW_IMAGE_DIR", os.path.join(config.PHOTO_DIR, "raw")
+        )
+    try:
+        img_copy = np.copy(image.image2cv(frame, False, False))
+        _save_queue.put_nowait(
+            {
+                "kind": "raw",
+                "img_cv": img_copy,
+                "shot_id": shot_id,
+                "photo_dir": photo_dir,
+            }
+        )
+    except queue.Full:
+        logger = logger_manager.logger
+        if logger:
+            logger.warning("[VISION-RAW] 存图队列已满，跳过本次纯原图保存")
+    except Exception as e:
+        logger = logger_manager.logger
+        if logger:
+            logger.error(f"[VISION-RAW] 复制纯原图失败: {e}")
+
+
 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, force_save=False):