Update scene_analyzer.py

scene_analyzer.py

@@ -72,10 +72,9 @@ class SceneAnalyzer:
         """
         Use new implement
         """
-        # 獲取功能區域信息（如果需要的話）
+        # get the functional zones info
         functional_zones = self.spatial_analyzer._identify_functional_zones(detected_objects, scene_type)
 
-        # 使用增強的場景描述生成器
         return self.generate_scene_description(
             scene_type,
             detected_objects,
@@ -123,7 +122,7 @@ class SceneAnalyzer:
                 "regions": {},
                 "possible_activities": [],
                 "safety_concerns": [],
-                "lighting_conditions": lighting_info or {"time_of_day": "unknown", "confidence": 0.0}
+                "lighting_conditions": lighting_info or {"time_of_day": "unknown", "confidence": 0}
             }
 
         # Get class names from detection result if not already set
@@ -149,7 +148,7 @@ class SceneAnalyzer:
                 "regions": {},
                 "possible_activities": [],
                 "safety_concerns": [],
-                "lighting_conditions": lighting_info or {"time_of_day": "unknown", "confidence": 0.0}
+                "lighting_conditions": lighting_info or {"time_of_day": "unknown", "confidence": 0}
             }
 
         # Analyze object distribution in regions
@@ -257,7 +256,7 @@ class SceneAnalyzer:
         # 添加 CLIP 特定的結果（新增）
         if clip_analysis and "error" not in clip_analysis:
             result["clip_analysis"] = {
-                "top_scene": clip_analysis.get("top_scene", ("unknown", 0.0)),
+                "top_scene": clip_analysis.get("top_scene", ("unknown", 0)),
                 "cultural_analysis": clip_analysis.get("cultural_analysis", {})
             }
 
@@ -375,15 +374,15 @@ class SceneAnalyzer:
 
         for scene_type in all_scene_types:
             # 獲取兩個模型的分數
-            yolo_score = yolo_scene_scores.get(scene_type, 0.0)
-            clip_score = clip_scene_scores.get(scene_type, 0.0)
+            yolo_score = yolo_scene_scores.get(scene_type, 0)
+            clip_score = clip_scene_scores.get(scene_type, 0)
 
             # 設置基本權重
-            yolo_weight = 0.7  # YOLO 提供更詳細的物體資訊
-            clip_weight = 0.3  # CLIP 提供更好的整體場景理解
+            yolo_weight = 0.7  # YOLO 可提供比較好的物體資訊
+            clip_weight = 0.3  # CLIP 強項是理解整體的場景關係
 
             # 對特定類型場景調整權重
-            # 文化特定場景或具有特殊布局的場景，CLIP 可能有優勢
+            # 文化特定場景或具有特殊布局的場景，CLIP可能比較能理解
             if any(keyword in scene_type for keyword in ["asian", "cultural", "aerial"]):
                 yolo_weight = 0.3
                 clip_weight = 0.7
@@ -393,7 +392,7 @@ class SceneAnalyzer:
                 yolo_weight = 0.8
                 clip_weight = 0.2
             elif scene_type == "beach_water_recreation":
-                yolo_weight = 0.8  # 衝浪板等特定物品的檢測非常重要
+                yolo_weight = 0.8  # 衝浪板等特定物品的檢測
                 clip_weight = 0.2
             elif scene_type == "sports_venue":
                 yolo_weight = 0.7