Scholarlink

This study proposes an underwater image instance segmentation algorithm based on YOLOv9 and an improved non-local module (non-local-s). By introducing the improved non-local module, the algorithm enhances the model’s ability to capture long-distance dependencies within images, thereby improving the accuracy of instance segmentation. Experimental results on the UIIS (Underwater Object Instance Segmentation) dataset demonstrate that the proposed algorithm significantly improves both accuracy and performance compared to existing state-of-the-art methods.

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

YOLOv9-N: An Underwater Image Instance Segmentation Algorithm Based on YOLOv9 and Improved Non-local Modules

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Shunong Tang,
  • Xuebo Jin,
  • Yuting Bai,
  • Tingli Su,
  • Jianlei Kong

摘要

This study proposes an underwater image instance segmentation algorithm based on YOLOv9 and an improved non-local module (non-local-s). By introducing the improved non-local module, the algorithm enhances the model’s ability to capture long-distance dependencies within images, thereby improving the accuracy of instance segmentation. Experimental results on the UIIS (Underwater Object Instance Segmentation) dataset demonstrate that the proposed algorithm significantly improves both accuracy and performance compared to existing state-of-the-art methods.