Saved in:
Bibliographic Details
Main Authors: He, Bo, Li, Hengduo, Jang, Young Kyun, Jia, Menglin, Cao, Xuefei, Shah, Ashish, Shrivastava, Abhinav, Lim, Ser-Nam
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2404.05726
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913327223930880
author He, Bo
Li, Hengduo
Jang, Young Kyun
Jia, Menglin
Cao, Xuefei
Shah, Ashish
Shrivastava, Abhinav
Lim, Ser-Nam
author_facet He, Bo
Li, Hengduo
Jang, Young Kyun
Jia, Menglin
Cao, Xuefei
Shah, Ashish
Shrivastava, Abhinav
Lim, Ser-Nam
contents With the success of large language models (LLMs), integrating the vision model into LLMs to build vision-language foundation models has gained much more interest recently. However, existing LLM-based large multimodal models (e.g., Video-LLaMA, VideoChat) can only take in a limited number of frames for short video understanding. In this study, we mainly focus on designing an efficient and effective model for long-term video understanding. Instead of trying to process more frames simultaneously like most existing work, we propose to process videos in an online manner and store past video information in a memory bank. This allows our model to reference historical video content for long-term analysis without exceeding LLMs' context length constraints or GPU memory limits. Our memory bank can be seamlessly integrated into current multimodal LLMs in an off-the-shelf manner. We conduct extensive experiments on various video understanding tasks, such as long-video understanding, video question answering, and video captioning, and our model can achieve state-of-the-art performances across multiple datasets. Code available at https://boheumd.github.io/MA-LMM/.
format Preprint
id arxiv_https___arxiv_org_abs_2404_05726
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MA-LMM: Memory-Augmented Large Multimodal Model for Long-Term Video Understanding
He, Bo
Li, Hengduo
Jang, Young Kyun
Jia, Menglin
Cao, Xuefei
Shah, Ashish
Shrivastava, Abhinav
Lim, Ser-Nam
Computer Vision and Pattern Recognition
With the success of large language models (LLMs), integrating the vision model into LLMs to build vision-language foundation models has gained much more interest recently. However, existing LLM-based large multimodal models (e.g., Video-LLaMA, VideoChat) can only take in a limited number of frames for short video understanding. In this study, we mainly focus on designing an efficient and effective model for long-term video understanding. Instead of trying to process more frames simultaneously like most existing work, we propose to process videos in an online manner and store past video information in a memory bank. This allows our model to reference historical video content for long-term analysis without exceeding LLMs' context length constraints or GPU memory limits. Our memory bank can be seamlessly integrated into current multimodal LLMs in an off-the-shelf manner. We conduct extensive experiments on various video understanding tasks, such as long-video understanding, video question answering, and video captioning, and our model can achieve state-of-the-art performances across multiple datasets. Code available at https://boheumd.github.io/MA-LMM/.
title MA-LMM: Memory-Augmented Large Multimodal Model for Long-Term Video Understanding
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2404.05726