Update app.py

app.py

@@ -20,7 +20,7 @@ logger = logging.getLogger(__name__)
 
 
 # 下載YOLOv8預訓練模型
-model_yolo = YOLO('yolov8n.pt')  # 使用 YOLOv8 預訓練模型
+model_yolo = YOLO('yolov8s.pt')  # 使用 YOLOv8 預訓練模型
 
 
 dog_breeds = ["Afghan_Hound", "African_Hunting_Dog", "Airedale", "American_Staffordshire_Terrier", 
@@ -229,32 +229,32 @@ async def detect_multiple_dogs(image, conf_threshold=0.15, iou_threshold=0.3):
             confidence = box.conf.item()
             boxes.append((xyxy, confidence))
 
-    # 如果沒有檢測到任何狗
     if not boxes:
         dogs.append((image, 1.0, [0, 0, image.width, image.height]))
     else:
-        nms_boxes = non_max_suppression(boxes, iou_threshold)
+        # 新增框的面積過濾條件，避免太小的框
+        valid_boxes = [box for box in boxes if (box[0][2] - box[0][0]) * (box[0][3] - box[0][1]) > 0.1 * image.width * image.height]
+        
+        nms_boxes = non_max_suppression(valid_boxes, iou_threshold)
         
-        # 進一步優化處理重疊框邏輯
         for box, confidence in nms_boxes:
             x1, y1, x2, y2 = box
             w, h = x2 - x1, y2 - y1
             
-            # 計算高度和寬度的比率，如果比例異常，則認定為重疊框需要拆分
-            aspect_ratio = h / w if w != 0 else 1
-            if aspect_ratio > 1.5 or aspect_ratio < 0.5:
-                # 假設重疊度過高，可以進一步裁切框
-                x1 = max(0, x1 - w * 0.05)
-                y1 = max(0, y1 - h * 0.05)
-                x2 = min(image.width, x2 + w * 0.05)
-                y2 = min(image.height, y2 + h * 0.05)
+            # 調整框的位置，處理重疊框問題
+            if w * h < 0.2 * image.width * image.height and confidence < 0.2:
+                continue  # 跳過信心分數過低的框
+
+            # 根據框的大小動態調整信心門檻
+            if w * h < 0.05 * image.width * image.height:
+                continue  # 過小的框直接跳過
             
             cropped_image = image.crop((x1, y1, x2, y2))
             dogs.append((cropped_image, confidence, [x1, y1, x2, y2]))
 
     return dogs
 
-# 增加一個優化的non_max_suppression版本
+
 def non_max_suppression(boxes, iou_threshold=0.3):
     keep = []
     boxes = sorted(boxes, key=lambda x: x[1], reverse=True)  # 按信心分數排序
@@ -265,23 +265,23 @@ def non_max_suppression(boxes, iou_threshold=0.3):
         boxes = [box for box in boxes if calculate_iou(current[0], box[0]) < iou_threshold]
     return keep
 
+
 def calculate_iou(box1, box2):
     x1 = max(box1[0], box2[0])
     y1 = max(box1[1], box2[1])
     x2 = min(box1[2], box2[2])
     y2 = min(box1[3], box2[3])
     
-    # 計算交集面積
     intersection = max(0, x2 - x1) * max(0, y2 - y1)
     area1 = (box1[2] - box1[0]) * (box1[3] - box1[1])
     area2 = (box2[2] - box2[0]) * (box2[3] - box2[1])
     
-    # 計算IOU
     iou = intersection / float(area1 + area2 - intersection)
     return iou
 
 
 
+
 async def process_single_dog(image):
     top1_prob, topk_breeds, topk_probs_percent = await predict_single_dog(image)
     if top1_prob < 0.2: