MA-LMM: Memory-Augmented Large Multimodal Model for Long-Term Video Understanding
This addresses the challenge of long-term video analysis for AI systems, enabling better handling of extended video content without exceeding computational constraints, though it is incremental in building upon existing multimodal models.
The authors tackled the problem of long-term video understanding by proposing a memory-augmented large multimodal model that processes videos online and stores past information in a memory bank, achieving state-of-the-art performance across multiple datasets.
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/.