Scholarlink

<p>Shrimp aquaculture is a major contributor to the global seafood supply, with India playing a key role. However, disease outbreaks and suboptimal farm management continue to cause substantial production losses. This study presents an integrated IoT–machine learning framework for real-time monitoring, behavioural analysis, and early detection of acute stress in shrimp culture systems. Environmental data collected through IoT sensors were combined with computer vision and machine learning techniques to analyse shrimp behaviour under varying water quality conditions. A YOLOv5 deep learning model enabled reliable underwater shrimp detection and tracking, achieving 84% detection accuracy. Behavioural changes driven by environmental stressors were predicted using machine learning models, with Decision Tree and Naïve Bayes classifiers achieving accuracies of 92% for pH-related responses and 88% for dissolved oxygen-related responses, respectively. Predicted behavioural anomalies were further validated through differential hemocyte count analysis, establishing a physiological link between environmental stress and observed behaviour. The proposed framework demonstrates a practical, data-driven approach for early stress detection, supporting reduced mortality, improved farm management, and sustainable shrimp production aligned with the Sustainable Development Goals.</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

IoT and ML for identification and behavioural analysis in shrimp aquaculture

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月19日

查看全文
  • Liton Paul,
  • Vinod Kumar Yadav,
  • Tanushree Bhowmik,
  • Vidya Shree Bharti,
  • Arun Sharma,
  • Ashutosh Deo,
  • Arpita Sharma

摘要

Shrimp aquaculture is a major contributor to the global seafood supply, with India playing a key role. However, disease outbreaks and suboptimal farm management continue to cause substantial production losses. This study presents an integrated IoT–machine learning framework for real-time monitoring, behavioural analysis, and early detection of acute stress in shrimp culture systems. Environmental data collected through IoT sensors were combined with computer vision and machine learning techniques to analyse shrimp behaviour under varying water quality conditions. A YOLOv5 deep learning model enabled reliable underwater shrimp detection and tracking, achieving 84% detection accuracy. Behavioural changes driven by environmental stressors were predicted using machine learning models, with Decision Tree and Naïve Bayes classifiers achieving accuracies of 92% for pH-related responses and 88% for dissolved oxygen-related responses, respectively. Predicted behavioural anomalies were further validated through differential hemocyte count analysis, establishing a physiological link between environmental stress and observed behaviour. The proposed framework demonstrates a practical, data-driven approach for early stress detection, supporting reduced mortality, improved farm management, and sustainable shrimp production aligned with the Sustainable Development Goals.