wenhaochai/MovieChat
MovieChat is a multimodal LLM system that handles long video understanding by converting dense tokens to sparse memory for efficient processing on standard GPUs.
Not currently ranked — collecting fresh signals.
star history
MovieChat is a CVPR 2024 research paper and open-source implementation of a video understanding system that combines computer vision and language models. It uses a sparse memory mechanism to efficiently process videos with over 10K frames on a 24GB GPU, dramatically reducing memory overhead compared to traditional approaches. The system builds on LLaMA and includes a leaderboard (MovieChat-1K) for benchmarking long video understanding performance.
Frequently asked
- What is wenhaochai/MovieChat?
- MovieChat is a multimodal LLM system that handles long video understanding by converting dense tokens to sparse memory for efficient processing on standard GPUs.
- Is MovieChat open source?
- Yes — wenhaochai/MovieChat is open source, released under the BSD-3-Clause license.
- What language is MovieChat written in?
- wenhaochai/MovieChat is primarily written in Python.
- How popular is MovieChat?
- wenhaochai/MovieChat has 704 stars on GitHub.
- Where can I find MovieChat?
- wenhaochai/MovieChat is on GitHub at https://github.com/wenhaochai/MovieChat.