'commit'

Apps/XinDianTu/Kit.py

@@ -204,8 +204,8 @@ def detect_rabbit_ad_close(image_path, debug_dir=None):
                         cX = int(M["m10"] / M["m00"]) + roi_x1
                         cY = int(M["m01"] / M["m00"]) + roi_y1
                         
-                        norm_x = int(cX / w * 1000)
-                        norm_y = int(cY / h * 1000)
+                        norm_x = cX / w
+                        norm_y = cY / h
                         
                         # 避免重复
                         if not any(abs(cX - c[0]) < 15 and abs(cY - c[1]) < 15 for c in candidates):
@@ -220,17 +220,23 @@ def detect_rabbit_ad_close(image_path, debug_dir=None):
         has_x = c[5]
         # 基础分：如果有 X，大幅加分
         score = 1000 if has_x else 0
-        # 距离分：越靠近预期的 (93, 830) 分越高
-        dist = np.sqrt((c[3] - 93)**2 + (c[4] - 830)**2)
-        score -= dist * 2
+        # 距离分：越靠近预期的 (0.094, 0.830) 分越高
+        dist = np.sqrt((c[3] - 0.094)**2 + (c[4] - 0.830)**2)
+        score -= dist * 2000  # 归一化后距离变小，需加大权重
         # 面积分：理想面积在 500-1500 之间
         if 500 < c[2] < 1500: score += 200
         return score
 
     candidates.sort(key=score_candidate, reverse=True)
     best = candidates[0]
+    best_score = score_candidate(best)
     
-    logger.info(f"CV detected rabbit ad close button at Norm({best[3]}, {best[4]}) with score {score_candidate(best):.2f}")
+    logger.info(f"CV detected rabbit ad close button at Norm({best[3]:.3f}, {best[4]:.3f}) with score {best_score:.2f}")
+    
+    # 【优化】如果得分太低 (低于 850)，说明误判概率较大，不予返回
+    if best_score < 850:
+        logger.info(f"Score {best_score:.2f} is below threshold 850, ignoring candidate.")
+        return None
     
     if debug_dir:
         os.makedirs(debug_dir, exist_ok=True)
@@ -749,13 +755,19 @@ def crop_cards_from_image(img_path, output_dir=None, save_debug=True):
             # 使用 Median 比 Max 更稳健，避免被单个超高卡片(如广告)带偏
             threshold_h = median_h * 0.70
 
-            for (y1, y2), h in zip(temp_valid_segments, heights):
-                if h < threshold_h:
+            for (y1, y2), card_h in zip(temp_valid_segments, heights):
+                # 【优化】过滤顶部非场站区域 (例如会员图标、搜索栏等)
+                # 场站列表通常在屏幕 35% 高度以后
+                if y1 / h < 0.35:
+                    logger.info(f"  Filtering out segment Y={y1}-{y2} because it's too high up (Top {y1/h:.2f} < 0.35).")
+                    continue
+
+                if card_h < threshold_h:
                     logger.info(
-                        f"  Filtering out segment Y={y1}-{y2} (H={h}) because it's too short (Threshold={threshold_h:.1f}).")
+                        f"  Filtering out segment Y={y1}-{y2} (H={card_h}) because it's too short (Threshold={threshold_h:.1f}).")
                 else:
                     final_cards.append((y1, y2, final_x1, final_x2))
-                    logger.info(f"  Card: Y={y1}-{y2}, X={final_x1}-{final_x2}, H={h}")
+                    logger.info(f"  Card: Y={y1}-{y2}, X={final_x1}-{final_x2}, H={card_h}")
 
     # 4. 保存结果
     if output_dir is None:
@@ -812,6 +824,12 @@ def crop_cards_from_image(img_path, output_dir=None, save_debug=True):
         card_info = {
             "id": idx + 1,
             "rect": [x1, y1, x2, y2],
+            "bounds_norm": {
+                "left": x1 / w,
+                "top": y1 / h,
+                "right": x2 / w,
+                "bottom": y2 / h
+            },
             "click_point": [click_x, click_y]
         }
         json_data["cards"].append(card_info)