Update app.py

app.py

@@ -1,80 +1,65 @@
-import os
-from operator import itemgetter
-import gradio as gr
-from mmaction.apis import init_recognizer, inference_recognizer
-
-CONFIG_FILE = 'demo/demo_configs/tsn_r50_1x1x8_video_infer.py'
-CHECKPOINT_FILE = 'checkpoints/tsn_r50_1x1x3_100e_kinetics400_rgb_20200614-e508be42.pth'
-LABEL_FILE = 'tools/data/kinetics/label_map_k400.txt'
-
-def load_labels(path):
-    if os.path.exists(path):
-        with open(path, 'r') as f:
-            return [x.strip() for x in f if x.strip()]
-    return None
-
-def build_model():
-    if not os.path.exists(CONFIG_FILE):
-        raise FileNotFoundError(f'Config not found at {CONFIG_FILE}')
-    if not os.path.exists(CHECKPOINT_FILE):
-        raise FileNotFoundError(f'Checkpoint not found at {CHECKPOINT_FILE}')
-    return init_recognizer(CONFIG_FILE, CHECKPOINT_FILE, device='cpu')
-
-print('Initializing model...')
-try:
-    model = build_model()
-    print('✅ Model loaded successfully!')
-except Exception as e:
-    print(f'❌ Error loading model: {e}')
-    model = None
-
-labels = load_labels(LABEL_FILE)
-def _resolve_video_path(video_input):
-    if isinstance(video_input, str):
-        return video_input
-    if isinstance(video_input, dict):
-        for key in ('name', 'path', 'video', 'file'):
-            val = video_input.get(key)
-            if isinstance(val, str) and os.path.exists(val):
-                return val
-    return video_input
-
-def analyze_video(video_input):
-    try:
-        if video_input is None:
-            return 'Please upload a video file.'
-        if model is None:
-            return '⚠️ Model not loaded. Check logs for details.'
-        video_path = _resolve_video_path(video_input)
-        if not isinstance(video_path, str) or not os.path.exists(video_path):
-            return '❌ Could not resolve uploaded video path.'
-
-        result = inference_recognizer(model, video_path)
-
-        pred_scores = result.pred_score.tolist()
-        score_sorted = sorted(zip(range(len(pred_scores)), pred_scores), key=itemgetter(1), reverse=True)
-        top5 = score_sorted[:5]
-
-        lines = []
-        for idx, score in top5:
-            name = labels[idx] if labels and idx < len(labels) else f'class_{idx}'
-            lines.append(f'{name}:  {score}')
-        return '\n'.join(lines)
-    except Exception as e:
-        return f'❌ Error processing video: {str(e)}'
-
-demo = gr.Interface(
-    fn=analyze_video,
-    inputs=gr.Video(label='Upload Video', height=300),
-    outputs=gr.Textbox(label='Analysis Results', lines=12),
-    title='🎬 GenVidBench - TSN (MMAction2)',
-    description='Upload a video. Inference uses TSN R50 on Kinetics-400.',
-    cache_examples=False,
-    flagging_mode='never'
-)
-
-if __name__ == '__main__':
-    demo.launch()
-
-
-
+import os
+import gradio as gr
+from operator import itemgetter
+from mmaction.apis import init_recognizer, inference_recognizer
+
+# --- Config & Checkpoint ---
+config_file = "demo/demo_configs/tsn_r50_1x1x8_video_infer.py"
+checkpoint_file = "checkpoints/tsn_r50_8xb32-1x1x8-100e_kinetics400-rgb_20220818-2692d16c.pth"
+label_file = "tools/data/kinetics/label_map_k400.txt"
+
+# --- Initialize model ---
+device = "cuda" if os.path.exists("/dev/nvidia0") else "cpu"
+print(f"🚀 Initializing TSN Model on {device}...")
+model = init_recognizer(config_file, checkpoint_file, device=device)
+
+# Load labels
+with open(label_file) as f:
+    labels = [x.strip() for x in f.readlines()]
+
+def analyze_video(video_path):
+    """Run action recognition on uploaded video"""
+    try:
+        if video_path is None:
+            return "❌ Please upload a video file."
+
+        # Inference
+        results = inference_recognizer(model, video_path)
+
+        # Extract top-5 results
+        pred_scores = results.pred_score.tolist()
+        score_tuples = tuple(zip(range(len(pred_scores)), pred_scores))
+        score_sorted = sorted(score_tuples, key=itemgetter(1), reverse=True)
+        top5 = score_sorted[:5]
+
+        # Format results
+        lines = []
+        for idx, score in top5:
+            lines.append(f"{labels[idx]}: {score:.4f}")
+
+        return "\n".join(lines)
+
+    except Exception as e:
+        return f"❌ Error: {str(e)}"
+
+# --- Gradio UI ---
+demo = gr.Interface(
+    fn=analyze_video,
+    inputs=gr.Video(label="Upload Video", height=300, type="filepath"),
+    outputs=gr.Textbox(label="Top-5 Predictions", lines=10),
+    title="🎬 GenVidBench - TSN Action Recognition",
+    description="""
+    Upload a video and run **TSN (Temporal Segment Networks, ResNet-50 backbone)**
+    trained on **Kinetics-400**.  
+    Model: `tsn_r50_8xb32-1x1x8-100e_kinetics400-rgb`  
+    Benchmark accuracy ~80% (GenVidBench).
+    """,
+    examples=[["demo/demo.mp4"]] if os.path.exists("demo/demo.mp4") else None,
+    cache_examples=False,
+    theme=gr.themes.Soft(),
+    allow_flagging="never"
+)
+
+if __name__ == "__main__":
+    print("🌟 Starting GenVidBench TSN Demo...")
+    demo.launch()