Gait recognition based on GaitPart and multimodal feature fusion

Pengpeng Chen; Chuanping Hu

doi:10.1117/12.3010587

1 December 2023 Gait recognition based on GaitPart and multimodal feature fusion

Pengpeng Chen, Chuanping Hu

Proceedings Volume 12940, Third International Conference on Control and Intelligent Robotics (ICCIR 2023); 129403C (2023) https://doi.org/10.1117/12.3010587
Event: Third International Conference on Control and Intelligent Robotics (ICCIR 2023), 2023, Sipsongpanna, China

Abstract

A new multimodal feature fusion representation method is proposed to address the problem of low accuracy in gait recognition under single-modal conditions. The contour features and skeletal features are obtained using background subtraction and human pose estimator, respectively. The two types of features are fused in an embedded manner to generate contour-skeleton feature images. The resulting skeleton-contour sequences are horizontally segmented and processed individually. Finally, the extracted feature vectors are mapped to a metric space using fully connected layers. Furthermore, in the existing GaitPart algorithm, the triplet loss function is employed for distance metric learning during network training. However, it suffers from slow convergence and unstable performance. To enhance the discriminative ability of the network, we propose a joint loss function that combines the triplet loss and softmax loss during the training process. Finally, extensive experiments are conducted on the CASIA-B dataset to evaluate the effectiveness of our proposed approach.

(2023) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Pengpeng Chen and Chuanping Hu "Gait recognition based on GaitPart and multimodal feature fusion", Proc. SPIE 12940, Third International Conference on Control and Intelligent Robotics (ICCIR 2023), 129403C (1 December 2023); https://doi.org/10.1117/12.3010587

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available