Making person search enjoy the merits of person re-identification

Person search is an extended task of person re-identification (Re-ID). However, most existing one-step works do not study how to employ Re-ID models improve the search. To address this issue, we propose a Teacher-guided Disentangling Network (TDN) make enjoy merits research. The proposed TDN can significantly boost performance by transferring advanced knowledge model. In TDN, for better transfer from teacher model model, design new base framework partially disentangling two subtasks. Besides, Knowledge Transfer Bridge module bridge scale gap caused different input formats between and Moreover, also Ranking with Context Persons strategy exploit context information in panoramic images ranking. Experiments on public datasets demonstrate favorable method.