Discriminative regularized metric learning for person re-identification
Venice Erin Liong, Yongxin Ge, Jiwen Lu · 2015
Person re-identification aims to match people across non-overlapping cameras, and recent advances have shown that metric learning is an effective technique for person re-identification. However, most existing metric learning methods suffer from the small sample size (SSS) problem due to the limited amount of labeled training samples. In this paper, we propose a new discriminative regularized metric learning (DRML) method for person re-identification. Specifically, we exploit discriminative information of training samples to regulate the eigenvalues of the intra-class and inter-class covariance matrices so that the distance metric estimated is less biased. Experimental results on three widely used datasets validate the effectiveness of our proposed method for person re-identification.