Scholarlink

<p>This study presents an interactive AI-based crop and pest management system integrated into a mobile application, designed to enhance sustainable agricultural productivity through the use of advanced deep learning techniques. The proposed framework comprises two core modules: (i) Crop Recommendation and (ii) Pest Detection. The Crop Recommendation module utilizes a Convolutional Neural Network (CNN) built on transfer learning to process agronomic inputs such as macronutrient levels (Nitrogen, Phosphorus, Potassium), soil pH, and climatic parameters (Temperature, Humidity, Rainfall). This model is optimized with a learning rate of 0.0001, batch size of 16, and trained over 100 epochs. Following crop recommendation, the pest detection and recommendation module is executed using the IP102 dataset to ensure targeted pest identification and treatment. The Pest Detection module leverages EfficientNet B4, employing a learning rate of 0.001, batch size of 64, and trained for 11 epochs with ReLU and Softmax activations, a dropout rate of 0.4, and weight decay of 0.005. It processes high-resolution images (1024 × 1024 pixels) of rice crops annotated with disease types and severity to detect pest species and recommend targeted pesticide interventions. The system achieved a training accuracy of 96.32% and testing accuracy of 82.54% with EfficientNet B4. Deployed on an Android platform, the application delivers real-time, field-level decision support for pest identification and crop advisory, enabling precision farming and improved resource utilization in sustainable agricultural systems also contributes significantly to global sustainability 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

An interactive AI-based crop and pest management system leveraging transfer learning for enhanced sustainable agriculture

  • 期刊论文

  • 发表日期：

    2025年5月14日

查看全文
  • Rakesh Kumar Ray,
  • Sujata Chakravarty,
  • Satyabrata Dash,
  • Sachi Nandan Mohanty,
  • Janjhyam Venkata Naga Ramesh

摘要

This study presents an interactive AI-based crop and pest management system integrated into a mobile application, designed to enhance sustainable agricultural productivity through the use of advanced deep learning techniques. The proposed framework comprises two core modules: (i) Crop Recommendation and (ii) Pest Detection. The Crop Recommendation module utilizes a Convolutional Neural Network (CNN) built on transfer learning to process agronomic inputs such as macronutrient levels (Nitrogen, Phosphorus, Potassium), soil pH, and climatic parameters (Temperature, Humidity, Rainfall). This model is optimized with a learning rate of 0.0001, batch size of 16, and trained over 100 epochs. Following crop recommendation, the pest detection and recommendation module is executed using the IP102 dataset to ensure targeted pest identification and treatment. The Pest Detection module leverages EfficientNet B4, employing a learning rate of 0.001, batch size of 64, and trained for 11 epochs with ReLU and Softmax activations, a dropout rate of 0.4, and weight decay of 0.005. It processes high-resolution images (1024 × 1024 pixels) of rice crops annotated with disease types and severity to detect pest species and recommend targeted pesticide interventions. The system achieved a training accuracy of 96.32% and testing accuracy of 82.54% with EfficientNet B4. Deployed on an Android platform, the application delivers real-time, field-level decision support for pest identification and crop advisory, enabling precision farming and improved resource utilization in sustainable agricultural systems also contributes significantly to global sustainability goals.