VLog aims to seek a new perspective on video-language understanding.
👇 Click the branch to see more instructions.
VLog (CVPR'25) | VLog-Agent | |
---|---|---|
TL;DR | Video Narration as Vocabulary | Video as Long Document |
![]() |
![]() |
|
Method | A novel, efficient video narrator (GPT2-based) with Narration Vocabulary via Generative Retrieval. | Given a video, we turn it into a textual document containing visual + audio info. By sending this doc to LLM, we can chat over the video! |
If you find our work helpful, please kindly consider citing our paper.
@misc{lin2025vlog,
title={VLog: Video-Language Models by Generative Retrieval of Narration Vocabulary},
author={Kevin Qinghong Lin and Mike Zheng Shou},
year={2025},
eprint={2503.09402},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2503.09402},
}