StreamBridge: Turning Your Offline Video Large Language Model into a Proactive Streaming Assistant

TL;DR

StreamBridge is a framework that enhances offline Video-LLMs with streaming capabilities, enabling real-time multi-turn understanding and proactive responses, supported by a new dataset and extensive experiments.

Contribution

It introduces a memory buffer with decay compression and a lightweight activation model, transforming offline Video-LLMs into effective streaming assistants.

Findings

Significantly improves streaming understanding in Video-LLMs

Outperforms proprietary models like GPT-4o and Gemini 1.5 Pro

Achieves competitive results on standard benchmarks

Abstract

We present StreamBridge, a simple yet effective framework that seamlessly transforms offline Video-LLMs into streaming-capable models. It addresses two fundamental challenges in adapting existing models into online scenarios: (1) limited capability for multi-turn real-time understanding, and (2) lack of proactive response mechanisms. Specifically, StreamBridge incorporates (1) a memory buffer combined with a round-decayed compression strategy, supporting long-context multi-turn interactions, and (2) a decoupled, lightweight activation model that can be effortlessly integrated into existing Video-LLMs, enabling continuous proactive responses. To further support StreamBridge, we construct Stream-IT, a large-scale dataset tailored for streaming video understanding, featuring interleaved video-text sequences and diverse instruction formats. Extensive experiments show that StreamBridge significantly improves the streaming understanding capabilities of offline Video-LLMs across various tasks, outperforming even proprietary models such as GPT-4o and Gemini 1.5 Pro. Simultaneously, it achieves competitive or superior performance on standard video understanding benchmarks.