Update module

app.py

@@ -91,6 +91,7 @@ def run_avse_inference(video_path, audio_path):
     vr = VideoReader(video_path, ctx=cpu(0))
     frames = vr.get_batch(list(range(len(vr)))).asnumpy()
     bg_frames = np.array([cv2.cvtColor(f, cv2.COLOR_RGB2GRAY) for f in frames]).astype(np.float32) / 255.0
+    bg_frames = np.expand_dims(bg_frames, axis=0)
 
     # Combine into input dict (match what model.enhance expects)
     data = {