Scholarlink

Underwater environments pose unique visual challenges such as noise, low resolution, and limited data. We propose a modified YOLOv8- x with a lightweight FasterNet backbone and a C2f-Em-Fast fusion block. Tests on public sonar datasets achieved 92.1% precision, 87.3% recall, and 86.4% mAP while reducing computation by over 80%, demonstrating real-time potential for sonar imaging.

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

Enhanced Sonar Scan Image Detection and Classification with a Modified YOLOv8-x Algorithm

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Estevao Siga,
  • Jing Zhang,
  • Félix Mérimé Bkangmo Kontchouo,
  • Bacem Saada,
  • Mingxiu Zhao,
  • TianChi Zhang

摘要

Underwater environments pose unique visual challenges such as noise, low resolution, and limited data. We propose a modified YOLOv8- x with a lightweight FasterNet backbone and a C2f-Em-Fast fusion block. Tests on public sonar datasets achieved 92.1% precision, 87.3% recall, and 86.4% mAP while reducing computation by over 80%, demonstrating real-time potential for sonar imaging.