Refactor object detection in app.py to use YOLOv8 model from Hugging Face, removing DiffusionPipeline. Update requirements.txt to include ultralytics and numpy, and adjust Gradio output type.

app.py

@@ -1,37 +1,24 @@
-import os
 import gradio as gr
-from diffusers import DiffusionPipeline
-import torch
-from huggingface_hub import login
+from ultralytics import YOLO
 
-
-login(token=os.getenv("HF_TOKEN"))
-
-# Initialize the pipeline
-pipe = DiffusionPipeline.from_pretrained("Lookingsoft-team/object_detection")
-if torch.cuda.is_available():
-    pipe = pipe.to("cuda")
+# Load YOLOv8 model from Hugging Face
+model = YOLO(
+    "https://huggingface.co/Lookingsoft-team/object_detection/resolve/main/yolov8n.pt"
+)
 
 
 def detect_objects(image):
-    if image is None:
-        return None
-
-    # Process the image through the pipeline
-    # Note: This is a placeholder - actual processing will depend on the model's specific requirements
-    results = pipe(image=image)
-
-    # Return the processed image with detections
-    return results.images[0]
+    results = model(image)
+    annotated_image = results[0].plot()  # Draw bounding boxes
+    return annotated_image
 
 
-# Create Gradio interface
 demo = gr.Interface(
     fn=detect_objects,
     inputs=gr.Image(type="pil"),
-    outputs=gr.Image(type="pil"),
+    outputs=gr.Image(type="numpy"),
     title="Object Detection",
-    description="Upload an image and the model will detect objects in it!",
+    description="Upload an image for object detection!",
 )
 
 if __name__ == "__main__":

requirements.txt

@@ -1,5 +1,5 @@
 gradio>=5.32.1
+ultralytics
 torch
-diffusers
-transformers
-accelerate 
+numpy
+Pillow