Think-Clip-Sample: Slow-Fast Frame Selection for Video Understanding

TL;DR

Think-Clip-Sample (TCS) is a training-free framework that improves long video understanding by adaptively balancing local details and global context, significantly boosting accuracy and efficiency across multiple models.

Contribution

TCS introduces a novel clip-level slow-fast sampling method combined with multi-query reasoning, enhancing long video comprehension without additional training.

Findings

Boosts accuracy up to 6.9% on benchmarks

Reduces inference time by 50% while maintaining accuracy

Effective across various multi-modal large language models

Abstract

Recent progress in multi-modal large language models (MLLMs) has significantly advanced video understanding. However, their performance on long-form videos remains limited by computational constraints and suboptimal frame selection. We present Think-Clip-Sample (TCS), a training-free framework that enhances long video understanding through two key components: (i) Multi-Query Reasoning, which generates multiple queries to capture complementary aspects of the question and video; and (ii) Clip-level Slow-Fast Sampling, which adaptively balances dense local details and sparse global context. Extensive experiments on MLVU, LongVideoBench, and VideoMME demonstrate that TCS consistently improves performance across different MLLMs, boosting up to 6.9% accuracy, and is capable of achieving comparable accuracy with 50% fewer inference time cost, highlighting both efficiency and efficacy of TCS on long video understanding.