基于Wi-Fi CSI的无监督域自适应伪装步态识别OA

Gait recognition,as a non-intrusive biometric technology,is widely used in security and smart home applications due to its ability to function without active user cooperation.However,a significant challenge faced by gait recognition systems in practice is the effect of disguise.When subjects alter their clothing or carry objects,the reliability of gait data is often compromised,making gait recognition difficult.To address this issue,this paper proposes a disguise gait recognition method based on Wi-Fi CSI(Channel State Information).The method introduces a novel data metric strategy,using pre-training to obtain pseudo-labels for disguised gait data,and employs matched filtering techniques to generate high-quality labeled training data pairs.Ultimately,it achieves data distribution alignment through unsupervised learning,overcoming the problem of distribution shift in disguised gait data.Experimental results show that the method significantly outperforms existing state-of-the-art gait recognition techniques in disguise scenarios.