Visible thermal person re-identification via multi-branch modality residual complementary learning

被引:1
|
作者
Chen, Long [1 ,2 ]
Sun, Rui [1 ,2 ,3 ]
Yu, Yiheng [1 ,2 ]
Du, Yun [1 ,2 ]
Zhang, Xudong [1 ,3 ]
机构
[1] Hefei Univ Technol, Sch Comp Sci & Informat Engn, 485 Danxia Rd, Hefei 230009, Peoples R China
[2] Hefei Univ Technol, Key Lab Ind Safety & Emergency Technol, Hefei 230009, Peoples R China
[3] Minist Educ Peoples Republ China, Key Lab Knowledge Engn Big Data, Hefei 230009, Peoples R China
基金
中国博士后科学基金; 中国国家自然科学基金;
关键词
Visible thermal person re-identification; Modality residual complementary learning; Multi-branch feature learning; Multi-branch constraint loss;
D O I
10.1016/j.imavis.2024.105201
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Visible-thermal Person Re-identification (VT Re-ID) is a challenging task in all-weather surveillance system. Existing methods concentrate on extracting the modality-shared features, ignoring the discriminative intermodality complementary features. To tackle this issue, we propose a multi-branch modality residual complementary learning method which consists of the modality residual complementary learning (MRCL) module and the multi-branch feature learning (MBFL) module. The MRCL module can be easily integrated into existing CNN baselines and drive the network to focus on both intra-modality and inter-modality information. On one hand, we adopt the basic two-stream network to obtain the intra-modality features, on the other hand, we capture the inter-modality complementary features within the residual image obtained by cross-modality correlation saliency erasing operation. To handle the intra-modality variations, we employ the MBFL module to capture local spatial features and local channel features, then integrate them with global features to achieve part-to-part and high-level semantic information matching. Finally, the discriminability and robustness of the ultimate representations are enhanced by multi-branch constraint loss learning. Extensive experiments on RegDB and SYSUMM01 datasets demonstrate the superiority of our proposed method compared with state-of-the-art methods.
引用
收藏
页数:11
相关论文
empty
未找到相关数据