【ReID】Learning Discriminative Features with Multiple Granularities for Person Re-identification

阅读2018 ACM Multimedia一篇著名SOTA，Learning Discriminative Features with Multiple Granularities for Person Re-Identification[1]，文章设计了一个结合多分支和表征度量双学习策略的切片网络，提取全局粗粒度和局部细粒度特征。通过多种细粒度特征的提取和学习来得到更好的具有判别力的表征，实验显

>>阅读原文<<