Create app.py

app.py

@@ -0,0 +1,104 @@
+import gradio as gr
+from ultralytics import YOLO
+import os
+import time
+from huggingface_hub import hf_hub_download
+
+# Setup model from Hugging Face
+REPO_ID = "ashen97/fabric-defect-yolo"
+MODEL_FILES = {
+    "YOLOv8n": "YOLOv8n.pt",
+    "YOLOv8s": "YOLOv8s.pt",
+    "YOLOv11": "YOLOv11.pt"
+}
+
+# Verify models
+print("Checking models...")
+MODEL_PATHS = {}
+
+try:
+    for display_name, filename in MODEL_FILES.items():
+        # Download model to local cache
+        cached_path = hf_hub_download(repo_id=REPO_ID, filename=filename)
+        MODEL_PATHS[display_name] = cached_path
+        print(f"✅ Loaded: {display_name}")
+except Exception as e:
+    print(f"⚠️ Note: Could not verify Hugging Face models ({e}).")
+
+# Prediction function
+def detect_defects(input_video, model_name, conf_level):
+    try:
+        if input_video is None:
+            raise gr.Error("Please upload a video first!")
+
+        # Create output folder
+        timestamp = int(time.time())
+        
+        output_dir = f"gradio_results_{timestamp}"
+        os.makedirs(output_dir, exist_ok=True)
+
+        # Load Model
+        if model_name in MODEL_PATHS:
+             model_path = MODEL_PATHS[model_name]
+        else:
+             raise gr.Error(f"Model file for {model_name} not found!")
+
+        model = YOLO(model_path)
+
+        # Run Prediction
+        print(f"Processing video with {model_name}...")
+        results = model.predict(
+            source = input_video,
+            save = True,
+            conf = conf_level,
+            project = output_dir,
+            name = "run",
+            verbose = False
+        )
+
+        # Find Output Video
+        save_dir = results[0].save_dir
+        files = os.listdir(save_dir)
+
+        # Verify video for .avi or .mp4
+        video_file = next((f for f in files if f.endswith(('.avi', '.mp4'))), None)
+
+        if not video_file:
+            raise gr.Error("YOLO finished but no video file was saved.")
+
+        full_video_path = os.path.join(save_dir, video_file)
+
+        # Convert to .mp4 for browser compatibility
+        output_mp4 = full_video_path.replace(".avi", "_converted.mp4")
+
+        # FFmpeg command
+        if not os.path.exists(output_mp4):
+            os.system(f"ffmpeg -y -loglevel panic -i '{full_video_path}' -vcodec libx264 '{output_mp4}'")
+
+        return output_mp4
+
+    # Show error in UI
+    except Exception as e:
+        print(f"Error Details: {str(e)}")
+        raise gr.Error(f"System Error: {str(e)}")
+
+# Build UI
+inputs = [
+    gr.Video(label="Input Video"),
+    gr.Dropdown(choices=list(MODEL_FILES.keys()), value=list(MODEL_FILES.keys())[0], label="Model Variant"),
+    gr.Slider(0.0, 1.0, value=0.25, step=0.05, label="Confidence Threshold")
+]
+
+# Run UI
+demo = gr.Interface(
+    fn = detect_defects,
+    inputs = inputs,
+    outputs = gr.Video(label="Predicted Output"),
+    title = "Fabric Defect Detection System",
+    description = "Upload a video to detect defects in real-time.",
+    theme = "soft"
+)
+
+# Launch
+if __name__ == "__main__":
+    demo.launch()