See More, Store Less: Memory-Efficient Resolution for Video Moment Retrieval

TL;DR

SMORE is a novel framework that improves memory efficiency in video moment retrieval by using query-guided encoding, importance modulation, and adaptive frame compression, achieving state-of-the-art results.

Contribution

Introduces SMORE, a memory-efficient video retrieval method that preserves key information through query-guided encoding and adaptive compression, outperforming existing approaches.

Findings

Achieves state-of-the-art performance on multiple benchmarks.

Effectively balances memory usage and information retention.

Enhances video understanding without exceeding memory budgets.

Abstract

Recent advances in Multimodal Large Language Models (MLLMs) have improved image recognition and reasoning, but video-related tasks remain challenging due to memory constraints from dense frame processing. Existing Video Moment Retrieval (VMR) methodologies rely on sparse frame sampling, risking potential information loss, especially in lengthy videos. We propose SMORE (See MORE, store less), a framework that enhances memory efficiency while maintaining high information resolution. SMORE (1) uses query-guided captions to encode semantics aligned with user intent, (2) applies query-aware importance modulation to highlight relevant segments, and (3) adaptively compresses frames to preserve key content while reducing redundancy. This enables efficient video understanding without exceeding memory budgets. Experimental validation reveals that SMORE achieves state-of-the-art performance on QVHighlights, Charades-STA, and ActivityNet-Captions benchmarks.