Update app.py

app.py

@@ -3,10 +3,10 @@ import cv2
 import numpy as np
 from ultralytics import YOLO
 
-# 加载YOLO模型
-model = YOLO('yolo11s-earth.pt')  # 加载你的模型
+# Load YOLO model
+model = YOLO('yolo11s-earth.pt')  # Load your model
 
-# 默认类别
+# Default classes
 default_classes = [
     'airplane', 'airport', 'baseballfield', 'basketballcourt', 'bridge',
     'chimney', 'dam', 'Expressway-Service-area', 'Expressway-toll-station',
@@ -16,41 +16,62 @@ default_classes = [
 ]
 
 def process_frame(frame, classes_input):
-    # 将输入的类别字符串转为列表
-    if classes_input:
+    # Process user input classes
+    if classes_input and classes_input.strip():
         classes_list = [cls.strip() for cls in classes_input.split(',')]
-        model.set_classes(classes_list)  # 设置模型的类别
+        # Validate classes_list
+        for cls in classes_list:
+            if not isinstance(cls, str):
+                print("Invalid class name:", cls)
+                continue
+        model.set_classes(classes_list)  # Set model classes
     else:
-        # 如果没有输入，则使用默认类别
+        # Use default classes if no input or input is empty
         model.set_classes(default_classes)
     
-    # 复制帧为可写数组
+    # Copy frame to a writable array
     frame = frame.copy()
     
-    # 转换图像格式
-    rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+    # Resize image to speed up processing (optional)
+    h, w = frame.shape[:2]
+    new_size = (640, int(h * (640 / w))) if w > h else (int(w * (640 / h)), 640)
+    resized_frame = cv2.resize(frame, new_size)
     
-    # 使用模型进行检测
-    results = model.predict(rgb_frame, imgsz=800)
+    # Convert image format
+    rgb_frame = cv2.cvtColor(resized_frame, cv2.COLOR_BGR2RGB)
     
-    # 绘制检测结果
+    # Use model for detection
+    results = model.predict(rgb_frame)
+    
+    # Draw detection results
     for result in results:
         boxes = result.boxes
         for box in boxes:
             x1, y1, x2, y2 = box.xyxy[0]
             conf = box.conf[0]
             cls = box.cls[0]
-            class_name = model.names[int(cls)]
+            try:
+                class_name = model.names[int(cls)]
+            except (IndexError, TypeError) as e:
+                print(f"Error accessing model.names: {e}")
+                class_name = "Unknown"  # Provide a default value
+            
+            # Adjust coordinates to original image size
+            x1 = int(x1 * w / new_size[0])
+            y1 = int(y1 * h / new_size[1])
+            x2 = int(x2 * w / new_size[0])
+            y2 = int(y2 * h / new_size[1])
             
-            # 绘制边界框和标签
-            cv2.rectangle(frame, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 2)
-            cv2.putText(frame, f'{class_name}:{conf:.2f}', (int(x1), int(y1) - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.9, (36, 255, 12), 2)
+            # Draw bounding box and label
+            cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)
+            cv2.putText(frame, f'{class_name}:{conf:.2f}', (x1, y1 - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.9, (36, 255, 12), 2)
     
     return frame
 
 def main():
-    # 创建Gradio界面
+    # Create Gradio interface
     with gr.Blocks() as demo:
+        gr.Markdown("# YOLO11s-Earth open vocabulary detection (DIOR finetuning)")
         with gr.Row():
             cam_input = gr.Image(type="numpy", sources=["webcam"], streaming=True, label="Webcam")
             classes_input = gr.Textbox(label="New classes (comma-separated)", placeholder="e.g.: airplane, airport, tennis court")
@@ -62,7 +83,7 @@ def main():
             outputs=output
         )
     
-    # 启动Gradio应用
+    # Launch Gradio app
     demo.launch()
 
 if __name__ == "__main__":