Tidy

app.py

@@ -65,10 +65,10 @@ def fn(video: gr.Video, actions: list[int]):
           f.write(image_encoded)
           yield f.name, None
           if toggle:
-            time.sleep(2)
+            time.sleep(0.5)
             toggle = False
       if i % config.classify_action_frame_step != 0 and i % config.detect_object_frame_step != 0:
-          time.sleep(0.4)
+          time.sleep(0.1)
     processed_clip = ImageSequenceClip(processed_frames, clip.fps)
     processed_clip.audio = clip.audio
     processed_clip.write_videofile(output, fps=clip.fps, audio_codec='aac', logger=None)

configuration.py

@@ -29,4 +29,4 @@ class Config:
   detect_object_frame_step = 5
   classify_action_frame_step = frame_step
   classify_action_num_frames = num_frames
-  yield_frame_steps = 1
+  yield_frame_steps = 4