Scholarlink

The aquaculture industry is an important pillar for China’s food security and economic development. However, with the global industrial intelligence development, the traditional aquaculture industry is now facing many challenges. Therefore, an efficient and accurate underwater fish target detection technology has become the key to improving the aquaculture efficiency and sustainability. Against this background, we propose an intelligent underwater fish detection system for freshwater and Marine environments - FisheryStar. This system is based on the YOLOv8 object detection algorithm, the efficient Feature Pyramid Network architecture (BiFPN) for multi-scale object detection, and the DeepSORT multi-object tracking algorithm. It can accurately identify underwater fish targets and their behavioral characteristics, and achieve three core functions: feeding optimization, disease early warning and growth rate detection. It solves the problems of high labour costs, low efficiency and poor precision in the traditional aquaculture industry, and also provides a reliable solution for building a smart fishery ecosystem. This article will explore in detail the architecture design, application scenarios and actual implementation effects of the “FisheryStar” system.

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

Underwater Fish Target Detection System Integrating Multi-scale Feature Enhancement and Multi-target Tracking Algorithms

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zicong Deng,
  • Lu Zeng,
  • Wenxi Xu,
  • Weichuan Ni

摘要

The aquaculture industry is an important pillar for China’s food security and economic development. However, with the global industrial intelligence development, the traditional aquaculture industry is now facing many challenges. Therefore, an efficient and accurate underwater fish target detection technology has become the key to improving the aquaculture efficiency and sustainability. Against this background, we propose an intelligent underwater fish detection system for freshwater and Marine environments - FisheryStar. This system is based on the YOLOv8 object detection algorithm, the efficient Feature Pyramid Network architecture (BiFPN) for multi-scale object detection, and the DeepSORT multi-object tracking algorithm. It can accurately identify underwater fish targets and their behavioral characteristics, and achieve three core functions: feeding optimization, disease early warning and growth rate detection. It solves the problems of high labour costs, low efficiency and poor precision in the traditional aquaculture industry, and also provides a reliable solution for building a smart fishery ecosystem. This article will explore in detail the architecture design, application scenarios and actual implementation effects of the “FisheryStar” system.