Update handler.py

handler.py

@@ -1,4 +1,4 @@
-from typing import Dict, List, Any
+from typing import Dict, Any
 from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
 from modelscope import snapshot_download
 from qwen_vl_utils import process_vision_info
@@ -7,9 +7,9 @@ import os
 import base64
 import io
 from PIL import Image
-import ffmpeg
 import logging
 import requests
+from moviepy.editor import VideoFileClip # For video frame extraction
 
 class EndpointHandler():
     def __init__(self, path=""):
@@ -98,31 +98,15 @@ class EndpointHandler():
         return image
 
     def _extract_video_frames(self, video_path, fps=1):
-        """Extracts frames from a video at the specified FPS."""
+        """Extracts frames from a video at the specified FPS using MoviePy."""
         try:
-            probe = ffmpeg.probe(video_path)
-            video_stream = next((stream for stream in probe['streams'] if stream['codec_type'] == 'video'), None)
-            if not video_stream:
-                logging.error(f"No video stream found in {video_path}")
-                return None
-
-            width = int(video_stream['width'])
-            height = int(video_stream['height'])
-
-            out, _ = (
-                ffmpeg
-                .input(video_path)
-                .filter('fps', fps=fps)
-                .output('pipe:', format='rawvideo', pix_fmt='rgb24')
-                .run(capture_stdout=True)
-            )
-            frames = []
-            for i in range(0, len(out), width * height * 3):
-                frame_data = out[i:i + width * height * 3]
-                frame = Image.frombytes('RGB', (width, height), frame_data)
-                frames.append(frame)
+            video = VideoFileClip(video_path)
+            frames = [
+                Image.fromarray(frame.astype('uint8'), 'RGB')
+                for frame in video.iter_frames(fps=fps)
+            ]
+            video.close()
             return frames
-
-        except ffmpeg.Error as e:
+        except Exception as e:
             logging.error(f"Error extracting video frames: {e}")
             return None