Face and Gait Recognition Based on Semi-supervised Learning

Abstract

The performance of the non-contact biometric recognition system is commonly poor when the labeled data set is small. To solve this problem, we perform the semi-supervised learning methods on face and gait to exploit the non-contact unlabeled biometric data. In the paper, the most important work is to apply co-training algorithm to the face and gait recognition system. Besides, we perform experiments on the database built by our group and obtain the results below: Co-training outperforms self-training in improving the performance of the biometric recognition system under same number of templates; Co-training uses fewer template than self-training (one vs. seven) to achieve best performance; Co-training suffers less impact than self-training from the different quality of initial templates.