Scholarlink

<p>Visible-Infrared Person Re-Identification (VI-ReID) plays a pivotal role in surveillance systems, enabling the accurate identification of individuals across varying times and locations. Traditional methods struggle in low-light conditions, which motivates our research. We introduce an Attention-Enhanced Feature Mapping Network (AEFMNet) that addresses both intra-modal and inter-modal discrepancies. Our AEFMNet employs an Attention-based Feature Fusion Module (AFFM) to enhance global feature representation and a GCN-based Feature Mapping Module (GFMM) to reduce cross-modal feature gaps. The proposed network is further strengthened by a Joint Training Algorithm (JTA) that integrates multi-scale local and global features, enhancing cross-modal matching accuracy. Our approach achieves advanced performance on three large-scale data sets, demonstrating its effectiveness and robustness.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Attention-enhanced feature mapping network for visible-infrared person re-identification

  • 期刊论文

  • 发表日期：

    2025年1月9日

查看全文
  • Shuaiyi Liu,
  • Ke Han

摘要

Visible-Infrared Person Re-Identification (VI-ReID) plays a pivotal role in surveillance systems, enabling the accurate identification of individuals across varying times and locations. Traditional methods struggle in low-light conditions, which motivates our research. We introduce an Attention-Enhanced Feature Mapping Network (AEFMNet) that addresses both intra-modal and inter-modal discrepancies. Our AEFMNet employs an Attention-based Feature Fusion Module (AFFM) to enhance global feature representation and a GCN-based Feature Mapping Module (GFMM) to reduce cross-modal feature gaps. The proposed network is further strengthened by a Joint Training Algorithm (JTA) that integrates multi-scale local and global features, enhancing cross-modal matching accuracy. Our approach achieves advanced performance on three large-scale data sets, demonstrating its effectiveness and robustness.