Deep Motion Prior for Weakly-Supervised Temporal Action Localization

TL;DR

This paper introduces a motion prior called motionness and a motion-guided loss to enhance weakly-supervised temporal action localization, significantly improving performance by better utilizing motion cues and addressing training loss issues.

Contribution

It proposes a novel motionness model based on optical flow and a motion-guided loss, which together improve action localization accuracy in weakly-supervised settings.

Findings

Achieves state-of-the-art results on THUMOS'14, ActivityNet v1.2, and v1.3.

Motionness effectively models action-relevant motion cues.

Motion-guided loss improves training and localization accuracy.

Abstract

Weakly-Supervised Temporal Action Localization (WSTAL) aims to localize actions in untrimmed videos with only video-level labels. Currently, most state-of-the-art WSTAL methods follow a Multi-Instance Learning (MIL) pipeline: producing snippet-level predictions first and then aggregating to the video-level prediction. However, we argue that existing methods have overlooked two important drawbacks: 1) inadequate use of motion information and 2) the incompatibility of prevailing cross-entropy training loss. In this paper, we analyze that the motion cues behind the optical flow features are complementary informative. Inspired by this, we propose to build a context-dependent motion prior, termed as motionness. Specifically, a motion graph is introduced to model motionness based on the local motion carrier (e.g., optical flow). In addition, to highlight more informative video snippets, a motion-guided loss is proposed to modulate the network training conditioned on motionness scores. Extensive ablation studies confirm that motionness efficaciously models action-of-interest, and the motion-guided loss leads to more accurate results. Besides, our motion-guided loss is a plug-and-play loss function and is applicable with existing WSTAL methods. Without loss of generality, based on the standard MIL pipeline, our method achieves new state-of-the-art performance on three challenging benchmarks, including THUMOS'14, ActivityNet v1.2 and v1.3.