GaitTAKE: Gait Recognition by Temporal Attention and Keypoint-guided Embedding

TL;DR

GaitTAKE introduces a novel gait recognition framework that leverages temporal attention and keypoint-guided embedding to improve accuracy by effectively fusing global, local, and pose features, setting new state-of-the-art results.

Contribution

The paper presents a new gait recognition method that incorporates learned temporal attention and keypoint-guided embedding for superior feature fusion.

Findings

Achieves 98.0% rank-1 accuracy on CASIA-B normal gait

Attains 97.5% accuracy on CASIA-B bag condition

Reaches 92.2% accuracy on CASIA-B coat condition

Abstract

Gait recognition, which refers to the recognition or identification of a person based on their body shape and walking styles, derived from video data captured from a distance, is widely used in crime prevention, forensic identification, and social security. However, to the best of our knowledge, most of the existing methods use appearance, posture and temporal feautures without considering a learned temporal attention mechanism for global and local information fusion. In this paper, we propose a novel gait recognition framework, called Temporal Attention and Keypoint-guided Embedding (GaitTAKE), which effectively fuses temporal-attention-based global and local appearance feature and temporal aggregated human pose feature. Experimental results show that our proposed method achieves a new SOTA in gait recognition with rank-1 accuracy of 98.0% (normal), 97.5% (bag) and 92.2% (coat) on the CASIA-B gait dataset; 90.4% accuracy on the OU-MVLP gait dataset.