GaitGCI: Generative Counterfactual Intervention for Gait Recognition

TL;DR

GaitGCI introduces a novel framework that enhances gait recognition by eliminating confounders and focusing on discriminative walking patterns through generative counterfactual interventions, achieving state-of-the-art results.

Contribution

The paper proposes GaitGCI, a new generative counterfactual intervention framework with CIL and DCDC modules, improving gait recognition by focusing on effective gait regions and reducing confounders.

Findings

Effectively focuses on discriminative gait regions

Plug-and-play with existing models for performance boost

Achieves state-of-the-art results in various scenarios

Abstract

Gait is one of the most promising biometrics that aims to identify pedestrians from their walking patterns. However, prevailing methods are susceptible to confounders, resulting in the networks hardly focusing on the regions that reflect effective walking patterns. To address this fundamental problem in gait recognition, we propose a Generative Counterfactual Intervention framework, dubbed GaitGCI, consisting of Counterfactual Intervention Learning (CIL) and Diversity-Constrained Dynamic Convolution (DCDC). CIL eliminates the impacts of confounders by maximizing the likelihood difference between factual/counterfactual attention while DCDC adaptively generates sample-wise factual/counterfactual attention to efficiently perceive the sample-wise properties. With matrix decomposition and diversity constraint, DCDC guarantees the model to be efficient and effective. Extensive experiments indicate that proposed GaitGCI: 1) could effectively focus on the discriminative and interpretable regions that reflect gait pattern; 2) is model-agnostic and could be plugged into existing models to improve performance with nearly no extra cost; 3) efficiently achieves state-of-the-art performance on arbitrary scenarios (in-the-lab and in-the-wild).