Efficient Temporal Extrapolation of Multimodal Large Language Models with Temporal Grounding Bridge
Yuxuan Wang, Yueqian Wang, Pengfei Wu, Jianxin Liang, Dongyan Zhao, Yang Liu, Zilong Zheng
TL;DR
This work addresses the bottlenecks of long-form video-language understanding by introducing Temporal Grounding Bridge (TGB), which enables efficient temporal grounding and context expansion through a low-dimensional, flow-based grounding backbone. It combines three innovations: (i) an efficient multi-span temporal grounding mechanism on low-dim temporal features, (ii) a length-extrapolation training paradigm with extrapolative position encoding to extend context windows, and (iii) a bootstrapping framework that plugs into pluggable MLLMs without requiring temporal grounding annotations. Across seven benchmarks and multiple MLLMs, TGB yields notable gains in long-form video QA and temporal grounding while maintaining computational efficiency, even enabling zero-shot generalization to longer sequences. The approach demonstrates strong generality and paves the way for practical video chat agents and scalable video-language systems, though it acknowledges remaining gaps in fine-grained temporal grounding.
Abstract
Despite progress in multimodal large language models (MLLMs), the challenge of interpreting long-form videos in response to linguistic queries persists, largely due to the inefficiency in temporal grounding and limited pre-trained context window size. In this work, we introduce Temporal Grounding Bridge (TGB), a novel framework that bootstraps MLLMs with advanced temporal grounding capabilities and broadens their contextual scope. Our framework significantly enhances the temporal capabilities of current MLLMs through three key innovations: an efficient multi-span temporal grounding algorithm applied to low-dimension temporal features projected from flow; a multimodal length extrapolation training paradigm that utilizes low-dimension temporal features to extend the training context window size; and a bootstrapping framework that bridges our model with pluggable MLLMs without requiring annotation. We validate TGB across seven video benchmarks and demonstrate substantial performance improvements compared with prior MLLMs. Notably, our model, initially trained on sequences of four frames, effectively handles sequences up to 16 longer without sacrificing performance, highlighting its scalability and effectiveness in real-world applications. Our code is publicly available at https://github.com/bigai-nlco/VideoTGB