Discriminative representation learning for person re-identification via multi-loss training.

Weilin Zhong,Tao Zhang,Linfeng Jiang,Jinsheng Ji,Zenghui Zhang,Huilin Xiong
DOI: https://doi.org/10.1016/j.jvcir.2019.06.001
IF: 2.887
2019-01-01
Journal of Visual Communication and Image Representation
Abstract:•The parametric center loss is extended into its non-parametric version.•A novel loss function is presented to further enlarge inter-class distance.•Multiple loss functions are combined to boost the performance of person re-id.
What problem does this paper attempt to address?