Scholarlink

In the evolving landscape of aquaculture, maintaining optimal fish health is pivotal for sustainability and productivity. Traditional health monitoring methods are labor-intensive and subject to human error, leading to inconsistencies in early disease detection and treatment. Utilizing the data obtained from James Cook University Singapore Aquaculture Laboratory consisting of videos of seabass (Dicentrarchus labrax) movements. We implemented a YOLOv8-based detection system to facilitate faster and more accurate detection of seabass under typical farming conditions. Our results demonstrate the effectiveness of this approach, achieving a mean average precision (mAP) at the 50th percentile intersection over union (IoU) threshold of 0.824 and a mAP between the 50th and 95th percentiles of 0.358. Future work will focus on expanding this detection system to perform real-time behavior analysis through fish trajectory tracking under various environmental conditions, enhancing early disease detection capabilities.

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

Enhancing Seabass Detection in Aquaculture: A Step Toward Automated Behavioral Analysis Using AI

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nitheesh Kumar Nallarasan,
  • FGhazalnaz Sharifonnasabi,
  • Nguyen Thanh Long,
  • Iman Makhdoom

摘要

In the evolving landscape of aquaculture, maintaining optimal fish health is pivotal for sustainability and productivity. Traditional health monitoring methods are labor-intensive and subject to human error, leading to inconsistencies in early disease detection and treatment. Utilizing the data obtained from James Cook University Singapore Aquaculture Laboratory consisting of videos of seabass (Dicentrarchus labrax) movements. We implemented a YOLOv8-based detection system to facilitate faster and more accurate detection of seabass under typical farming conditions. Our results demonstrate the effectiveness of this approach, achieving a mean average precision (mAP) at the 50th percentile intersection over union (IoU) threshold of 0.824 and a mAP between the 50th and 95th percentiles of 0.358. Future work will focus on expanding this detection system to perform real-time behavior analysis through fish trajectory tracking under various environmental conditions, enhancing early disease detection capabilities.