GaitSnippet: Gait Recognition Beyond Unordered Sets and Ordered Sequences

TL;DR

This paper introduces GaitSnippet, a novel gait recognition approach that models gait as a composition of snippets to better capture multi-scale temporal context, outperforming existing set and sequence methods.

Contribution

The paper proposes a new snippet-based gait recognition framework that effectively captures multi-scale temporal information, addressing limitations of prior set and sequence approaches.

Findings

Achieves 77.5% rank-1 accuracy on Gait3D dataset.

Achieves 81.7% rank-1 accuracy on GREW dataset.

Demonstrates effectiveness across four widely-used gait datasets.

Abstract

Recent advancements in gait recognition have significantly enhanced performance by treating silhouettes as either an unordered set or an ordered sequence. However, both set-based and sequence-based approaches exhibit notable limitations. Specifically, set-based methods tend to overlook short-range temporal context for individual frames, while sequence-based methods struggle to capture long-range temporal dependencies effectively. To address these challenges, we draw inspiration from human identification and propose a new perspective that conceptualizes human gait as a composition of individualized actions. Each action is represented by a series of frames, randomly selected from a continuous segment of the sequence, which we term a snippet. Fundamentally, the collection of snippets for a given sequence enables the incorporation of multi-scale temporal context, facilitating more comprehensive gait feature learning. Moreover, we introduce a non-trivial solution for snippet-based gait recognition, focusing on Snippet Sampling and Snippet Modeling as key components. Extensive experiments on four widely-used gait datasets validate the effectiveness of our proposed approach and, more importantly, highlight the potential of gait snippets. For instance, our method achieves the rank-1 accuracy of 77.5% on Gait3D and 81.7% on GREW using a 2D convolution-based backbone.