Scholarlink

<p>Technological developments are used in prawn farming to identify various diseases affecting prawns. This overcomes the drawbacks of conventional techniques, which require expert analysis and human examination, making them impractical for detecting the wide variety of bacterial and viral diseases that cause significant financial losses. The current machine learning models are capable of accurately detecting a wide range of diseases. To improve prawn disease detection in aquaculture, a new Hybrid Spatio-Temporal Transformer Network (HSTTN), combined with a dual optimization technique, is proposed. The suggested HSTTN uses temporal convolutional networks and spatial transformers to extract and examine temporal dynamics and spatial hierarchies, which are essential for precise disease progression analysis. To efficiently fuse this information and increase the accuracy of disease categorisation, a dynamic cross-attention technique is incorporated. To further fine-tune the classifier parameters, the model uses a hybrid optimisation technique that combines the Artificial Hummingbird Algorithm (AHA) and the Whale Optimisation Algorithm (WOA). Performance metrics such as accuracy, precision,recall,F1-score, specificity, and Matthew’s correlation coefficient are used in experiments to validate the proposed model. When compared to other Deep Learning (DL) techniques, such as the conventional Recurrent Neural Network (RNN), convolutional Neural Network (CNN), Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM) Networks, and Recurrent Capsule Networks, the suggested model performs better with precision of 97.4%, recall of 97.1%, f1-score of 97.2%, specificity of 97.9%, and accuracy of 98.1%.</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

A Hybrid Spatial Temporal Transformer Network Optimised for Accurate Disease Identification in Shrimp Producers

  • 期刊论文

  • 发表日期：

    2025年11月7日

查看全文
  • R. Rajalakshmi,
  • Vinod Kumar Shukla,
  • P. Sivakumar,
  • K. Manikanda Kumaran

摘要

Technological developments are used in prawn farming to identify various diseases affecting prawns. This overcomes the drawbacks of conventional techniques, which require expert analysis and human examination, making them impractical for detecting the wide variety of bacterial and viral diseases that cause significant financial losses. The current machine learning models are capable of accurately detecting a wide range of diseases. To improve prawn disease detection in aquaculture, a new Hybrid Spatio-Temporal Transformer Network (HSTTN), combined with a dual optimization technique, is proposed. The suggested HSTTN uses temporal convolutional networks and spatial transformers to extract and examine temporal dynamics and spatial hierarchies, which are essential for precise disease progression analysis. To efficiently fuse this information and increase the accuracy of disease categorisation, a dynamic cross-attention technique is incorporated. To further fine-tune the classifier parameters, the model uses a hybrid optimisation technique that combines the Artificial Hummingbird Algorithm (AHA) and the Whale Optimisation Algorithm (WOA). Performance metrics such as accuracy, precision,recall,F1-score, specificity, and Matthew’s correlation coefficient are used in experiments to validate the proposed model. When compared to other Deep Learning (DL) techniques, such as the conventional Recurrent Neural Network (RNN), convolutional Neural Network (CNN), Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM) Networks, and Recurrent Capsule Networks, the suggested model performs better with precision of 97.4%, recall of 97.1%, f1-score of 97.2%, specificity of 97.9%, and accuracy of 98.1%.