Project Gallery

The following open-source projects use PySceneDetect in their own code. They are listed for reference only: PySceneDetect does not endorse or maintain them, and each project is covered by its own license. Use them at your own risk.

If you know of a project that should be listed here, feel free to raise an issue with a link to where it uses PySceneDetect. Research papers that use PySceneDetect are listed on the Research & Literature page.

Datasets and Model Training

  • LTX-2 - the LTX-2 trainer's split_scenes.py script cuts training videos into scenes
  • Data-Juicer - the video_split_by_scene_mapper operator splits videos at scene changes, with a choice of detector
  • OneTrainer - the video tool finds cuts with AdaptiveDetector when extracting clips
  • VideoX-Fun - detects cuts in its video captioning pipeline

Video Generation

  • ViMax - splits generated transition videos into shots with split_video_ffmpeg
  • InfiniteTalk - detects shots in the input video, combining the content, adaptive, and threshold detectors

Agents, RAG, and Data Tools

  • ECC - measures the cut cadence of a reference video by sweeping ContentDetector thresholds
  • RAG-Anything - segments videos into scenes before indexing them
  • CAMEL - the VideoAnalysisToolkit extracts keyframes at scene changes
  • Skill Seekers - the video scraper finds scene boundaries, then extracts keyframes
  • OmAgent - splits videos into scenes for its video understanding agent
  • Pixeltable - provides scene_detect video functions, with a cookbook notebook
  • MassGen - the video extraction tool takes frames from each scene

Applications and Tools

  • Moondream - the content moderation recipe detects scenes before running the model
  • REAL Video Enhancer - includes a copy of ContentDetector to catch scene changes while rendering
  • movie-narrator - detects scenes as one step of its movie recap pipeline