Update app.py

app.py

@@ -1,5 +1,7 @@
 import gradio as gr
-from ultralytics import YOLOv10 
+import cv2
+import tempfile
+from ultralytics import YOLOv10
 import supervision as sv
 import spaces
 from huggingface_hub import hf_hub_download
@@ -30,28 +32,71 @@ category_dict = {
 }
 
 
-@spaces.GPU(duration=200)
-def yolov10_inference(image, model_id, image_size, conf_threshold, iou_threshold):
+def yolov10_inference(image, video, model_id, image_size, conf_threshold, iou_threshold):
     model_path = download_models(model_id)
     model = YOLOv10(model_path)
-    results = model(source=image, imgsz=image_size, iou=iou_threshold, conf=conf_threshold, verbose=False)[0]
-    detections = sv.Detections.from_ultralytics(results)
     
-    labels = [
-        f"{category_dict[class_id]} {confidence:.2f}"
-        for class_id, confidence in zip(detections.class_id, detections.confidence)
-    ]
-    annotated_image = box_annotator.annotate(image, detections=detections, labels=labels)
+    if image:
+        results = model(source=image, imgsz=image_size, iou=iou_threshold, conf=conf_threshold, verbose=False)[0]
+        detections = sv.Detections.from_ultralytics(results)
+        
+        labels = [
+            f"{category_dict[class_id]} {confidence:.2f}"
+            for class_id, confidence in zip(detections.class_id, detections.confidence)
+        ]
+        annotated_image = box_annotator.annotate(image, detections=detections, labels=labels)
+        return annotated_image[:, :, ::-1], None
+    else:
+        video_path = tempfile.mktemp(suffix=".webm")
+        with open(video_path, "wb") as f:
+            with open(video, "rb") as g:
+                f.write(g.read())
 
+        cap = cv2.VideoCapture(video_path)
+        fps = cap.get(cv2.CAP_PROP_FPS)
+        frame_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
+        frame_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
+
+        output_video_path = tempfile.mktemp(suffix=".webm")
+        out = cv2.VideoWriter(output_video_path, cv2.VideoWriter_fourcc(*'vp80'), fps, (frame_width, frame_height))
+
+        while cap.isOpened():
+            ret, frame = cap.read()
+            if not ret:
+                break
+
+            results = model(source=frame, imgsz=image_size, iou=iou_threshold, conf=conf_threshold, verbose=False)[0]
+            detections = sv.Detections.from_ultralytics(results)
+            
+            labels = [
+                f"{category_dict[class_id]} {confidence:.2f}"
+                for class_id, confidence in zip(detections.class_id, detections.confidence)
+            ]
+            annotated_frame = box_annotator.annotate(frame, detections=detections, labels=labels)
+            out.write(annotated_frame)
+
+        cap.release()
+        out.release()
+
+        return None, output_video_path
+
+
+def yolov10_inference_for_examples(image, model_id, image_size, conf_threshold, iou_threshold):
+    annotated_image, _ = yolov10_inference(image, None, model_id, image_size, conf_threshold, iou_threshold)
     return annotated_image
 
+
 def app():
     with gr.Blocks():
         with gr.Row():
             with gr.Column():
-                # image = gr.Image(type="pil", label="Image")
-                video = gr.Video(type="Video", visible=False)
-                
+                image = gr.Image(type="pil", label="Image", visible=True)
+                video = gr.Video(label="Video", visible=False)
+                input_type = gr.Radio(
+                    choices=["Image", "Video"],
+                    value="Image",
+                    label="Input Type",
+                )
                 model_id = gr.Dropdown(
                     label="Model",
                     choices=[
@@ -88,46 +133,54 @@ def app():
                 yolov10_infer = gr.Button(value="Detect Objects")
 
             with gr.Column():
-                output_image = gr.Image(type="pil", label="Annotated Image")
+                output_image = gr.Image(type="numpy", label="Annotated Image", visible=True)
                 output_video = gr.Video(label="Annotated Video", visible=False)
 
+        def update_visibility(input_type):
+            image_visibility = input_type == "Image"
+            return (
+                gr.update(visible=image_visibility),
+                gr.update(visible=not image_visibility),
+                gr.update(visible=image_visibility),
+                gr.update(visible=not image_visibility),
+            )
+
+        input_type.change(
+            fn=update_visibility,
+            inputs=[input_type],
+            outputs=[image, video, output_image, output_video],
+        )
+
+        def run_inference(image, video, model_id, image_size, conf_threshold, iou_threshold, input_type):
+            if input_type == "Image":
+                return yolov10_inference(image, None, model_id, image_size, conf_threshold, iou_threshold)
+            else:
+                return yolov10_inference(None, video, model_id, image_size, conf_threshold, iou_threshold)
+
         yolov10_infer.click(
-            fn=yolov10_inference,
-            inputs=[
-                image,
-                model_id,
-                image_size,
-                conf_threshold,
-                iou_threshold,
-            ],
-            outputs=[output_image],
+            fn=run_inference,
+            inputs=[image, video, model_id, image_size, conf_threshold, iou_threshold, input_type],
+            outputs=[output_image, output_video],
         )
 
         gr.Examples(
             examples=[
                 [
-                    "dog.jpeg",
-                    "yolov10x.pt",
+                    "ultralytics/assets/bus.jpg",
+                    "yolov10s.pt",
                     640,
                     0.25,
                     0.45,
                 ],
                 [
-                    "huggingface.jpg",
-                    "yolov10m.pt",
-                    640,
-                    0.25,
-                    0.45,
-                ],
-                [
-                    "zidane.jpg",
-                    "yolov10b.pt",
+                    "ultralytics/assets/zidane.jpg",
+                    "yolov10s.pt",
                     640,
                     0.25,
                     0.45,
                 ],
             ],
-            fn=yolov10_inference,
+            fn=yolov10_inference_for_examples,
             inputs=[
                 image,
                 model_id,
@@ -136,9 +189,10 @@ def app():
                 iou_threshold,
             ],
             outputs=[output_image],
-            cache_examples=True,
+            cache_examples='lazy',
         )
 
+
 gradio_app = gr.Blocks()
 with gradio_app:
     gr.HTML(
@@ -150,12 +204,12 @@ with gradio_app:
     gr.HTML(
         """
         <h3 style='text-align: center'>
-        Follow me for more!
-        <a href='https://twitter.com/kadirnar_ai' target='_blank'>Twitter</a> | <a href='https://github.com/kadirnar' target='_blank'>Github</a> | <a href='https://www.linkedin.com/in/kadir-nar/' target='_blank'>Linkedin</a>  | <a href='https://www.huggingface.co/kadirnar/' target='_blank'>HuggingFace</a>
+        <a href='https://arxiv.org/abs/2405.14458' target='_blank'>arXiv</a> | <a href='https://github.com/THU-MIG/yolov10' target='_blank'>github</a>
         </h3>
         """)
     with gr.Row():
         with gr.Column():
             app()
 
-gradio_app.launch(debug=True)
+if __name__ == '__main__':
+    gradio_app.launch()