Scholarlink

<p>To promote the technology of person re-identification (Re-ID) in intelligent video analysis, a new segmentation method of keypoint-based dynamic region partitioning (KDRP) and an improved adaptive average pooling layer list network (APLNet) are proposed in this work. The KDRP addresses the limitations of traditional stripe segmentation methods avoiding the influence of shooting angles and pedestrian postures. The APLNet integrates the adaptive average pooling layer list (AAPLL) module and the priority circle loss (P-circle loss) to solve the problem of inconsistent size of feature map and promote the model performance respectively. Experimental results on different datasets have validated the effectiveness of the proposed method.</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

Person re-identification method based on keypoint-based dynamic region partitioning and APLNet

  • 期刊论文

  • 发表日期：

    2026年4月10日

查看全文
  • Hongyi Wang,
  • Xirui Yang,
  • Xinjun Zhu,
  • Limei Song,
  • Yunpeng Li

摘要

To promote the technology of person re-identification (Re-ID) in intelligent video analysis, a new segmentation method of keypoint-based dynamic region partitioning (KDRP) and an improved adaptive average pooling layer list network (APLNet) are proposed in this work. The KDRP addresses the limitations of traditional stripe segmentation methods avoiding the influence of shooting angles and pedestrian postures. The APLNet integrates the adaptive average pooling layer list (AAPLL) module and the priority circle loss (P-circle loss) to solve the problem of inconsistent size of feature map and promote the model performance respectively. Experimental results on different datasets have validated the effectiveness of the proposed method.