You Can Ground Earlier than See: An Effective and Efficient Pipeline for Temporal Sentence Grounding in Compressed Videos

TL;DR

This paper introduces a novel compressed-domain approach for temporal sentence grounding in videos, utilizing raw bit-stream features to improve efficiency and effectiveness over traditional methods that rely on fully decoded frames.

Contribution

It proposes the Three-branch Compressed-domain Spatial-temporal Fusion (TCSF) framework that directly exploits compressed video features for temporal grounding, reducing computational complexity and latency.

Findings

TCSF outperforms state-of-the-art methods on three datasets.

The approach achieves higher accuracy with lower computational cost.

Utilizes I-frame, motion vector, and residual features for effective grounding.

Abstract

Given an untrimmed video, temporal sentence grounding (TSG) aims to locate a target moment semantically according to a sentence query. Although previous respectable works have made decent success, they only focus on high-level visual features extracted from the consecutive decoded frames and fail to handle the compressed videos for query modelling, suffering from insufficient representation capability and significant computational complexity during training and testing. In this paper, we pose a new setting, compressed-domain TSG, which directly utilizes compressed videos rather than fully-decompressed frames as the visual input. To handle the raw video bit-stream input, we propose a novel Three-branch Compressed-domain Spatial-temporal Fusion (TCSF) framework, which extracts and aggregates three kinds of low-level visual features (I-frame, motion vector and residual features) for effective and efficient grounding. Particularly, instead of encoding the whole decoded frames like previous works, we capture the appearance representation by only learning the I-frame feature to reduce delay or latency. Besides, we explore the motion information not only by learning the motion vector feature, but also by exploring the relations of neighboring frames via the residual feature. In this way, a three-branch spatial-temporal attention layer with an adaptive motion-appearance fusion module is further designed to extract and aggregate both appearance and motion information for the final grounding. Experiments on three challenging datasets shows that our TCSF achieves better performance than other state-of-the-art methods with lower complexity.