Update app.py

app.py

@@ -35,7 +35,7 @@ def read_video_opencv(file_path, indices):
         if success:
             # Convert BGR to RGB
             frames.append(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
-    return np.array(frames)
+    return frames
 
 
 def concatenate_frames(frames, clip_len):
@@ -71,6 +71,7 @@ def model_interface(uploaded_video, model_choice, activity):
     activities_list = [activity, "other"]
     processor = AutoProcessor.from_pretrained(model_choice)
     model = AutoModel.from_pretrained(model_choice)
+    
     inputs = processor(
         text=activities_list,
         videos=list(video),
@@ -119,4 +120,13 @@ iface = gr.Interface(
     live=False
 )
 
+
+inputs = processor(
+    text=activities_list,
+    videos=np.array(video),  # This line is changed.
+    return_tensors="pt",
+    padding=True,
+)
+
+
 iface.launch()