Where to Focus: Query-Modulated Multimodal Keyframe Selection for Long Video Understanding

TL;DR

The paper introduces Q-Gate, a dynamic, query-modulated framework for selecting keyframes in long videos, improving multimodal reasoning by intelligently routing modalities based on query intent without additional training.

Contribution

Q-Gate is a novel, training-free approach that dynamically allocates attention to different expert streams for keyframe selection based on query context, enhancing video understanding.

Findings

Q-Gate outperforms state-of-the-art methods on LongVideoBench and Video-MME.

It effectively suppresses modality-specific noise in multimodal video reasoning.

The approach is highly interpretable and adaptable across multiple MLLM backbones.

Abstract

Long video understanding remains a formidable challenge for Multimodal Large Language Models (MLLMs) due to the prohibitive computational cost of processing dense frame sequences. Prevailing solutions, which select a keyframe subset, typically rely on either a single visual-centric metric (e.g., CLIP similarity) or a static fusion of heuristic scores. This ``one-size-fits-all'' paradigm frequently fails: visual-only metrics are ineffective for plot-driven narrative queries, while indiscriminately incorporating textual scores introduces severe ``modal noise'' for purely visual tasks. To break this bottleneck, we propose Q-Gate, a plug-and-play and training-free framework that treats keyframe selection as a dynamic modality routing problem. We decouple the retrieval process into three lightweight expert streams: Visual Grounding for local details, Global Matching for scene semantics, and Contextual Alignment for subtitle-driven narratives. Crucially, Q-Gate introduces a Query-Modulated Gating Mechanism that leverages the in-context reasoning of an LLM to assess the query's intent and dynamically allocate attention weights across the experts. This mechanism intelligently activates necessary modalities while ``muting'' irrelevant ones, thereby maximizing the signal-to-noise ratio. Extensive experiments on LongVideoBench and Video-MME across multiple MLLM backbones demonstrate that Q-Gate substantially outperforms state-of-the-art baselines. By effectively suppressing modality-specific noise, it provides a robust, highly interpretable solution for scalable video reasoning.