A model-based gait recognition method with body pose and human prior knowledge

• Published in: Pattern recognition Vol. 98; p. 107069
• Main Authors: Liao, Rijun, Yu, Shiqi, An, Weizhi, Huang, Yongzhen
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.02.2020
• Subjects: Gait recognition; Human body pose; Spatio-temporal feature; Spatio-temporal feature; Human body pose; Gait recognition
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/j.patcog.2019.107069

•We propose a novel model-based gait recognition method, PoseGait, which exploits human pose as feature. The method can achieve high recognition rate despite the low dimensional feature (only 14 body joints).•We design dedicated features based on 3D pose information. We demonstrate experimentally the advantage of these features.•CNN nor RNN/LSTM can successfully extract spatio-temporal gait feature with the help of fusing two losses. We propose in this paper a novel model-based gait recognition method, PoseGait. Gait recognition is a challenging and attractive task in biometrics. Early approaches to gait recognition were mainly appearance-based. The appearance-based features are usually extracted from human body silhouettes, which are easy to compute and have shown to be efficient for recognition tasks. Nevertheless silhouettes shape is not invariant to changes in clothing, and can be subject to drastic variations, due to illumination changes or other external factors. An alternative to silhouette-based features are model-based features. However, they are very challenging to acquire especially for low image resolution. In contrast to previous approaches, our model PoseGait exploits human 3D pose estimated from images by Convolutional Neural Network as the input feature for gait recognition. The 3D pose, defined by the 3D coordinates of joints of the human body, is invariant to view changes and other external factors of variation. We design spatio-temporal features from the 3D pose to improve the recognition rate. Our method is evaluated on two large datasets, CASIA B and CASIA E. The experimental results show that the proposed method can achieve state-of-the-art performance and is robust to view and clothing variations.