Scholarlink

<p>In modern agriculture, integrating AI-oriented pest control and soil monitoring significantly improves crop productivity and sustainability. This article provides an innovative methodology for pest detection and control strategies and facilitates close-up soil monitoring with avoidance of soil-borne diseases in sugarcane fields by implementing integrated approaches of Deep Q-Network (DQN) and You Only Look Once Version 8 Medium (YOLOv8m) for object detection. The proposed work merges the features of YOLOv8m for effective pest identification with the existing dataset acquired from autonomous Soil Robots (ASR) designed for navigating and monitoring soil health conditions closely. YOLOv8m is used to recognize and categorize pests by giving a high precision rate in challenging environments. Here, drone-captured high-resolution pictures from the existing datasets of the sugarcane field have been used and processed by the proposed integrated models to acquire the best accuracy rate. This research mainly focuses on identifying major influencing pests like Early shoot borers, Black Bugs, Red Rust, Bacterial Blight, Stem Borers, Whiteflies, Red Rot, Leaf Spot, Mite Insect, and Crown Mealy Bugs based on the existing datasets. While, ASR with different sensors acquired data has been used and processed by the DQN algorithm to monitor the soil health based on the parameters of soil like pH, moisture, nutrient content, and temperature to avoid soil-borne diseases in sugarcane fields. The existing robots’ self-directed navigation abilities were offered by Reinforcement Learning, which allows the soil robots to adapt to various field conditions and prioritize regions with poor soil health conditions and pest detection hotspots. Experimental outcomes illustrate that the proposed integrated approach significantly enhances the accuracy of pest detection compared to traditional techniques by 98% and also provides detailed and close-up soil health monitoring. This existing data acquisition using drones and sensors allows better decision-making about pest control policies and soil handling measures, contributing to enhanced sugarcane crop health and productivity by the proposed integrated models. The proposed algorithm shows the feasibility of AI/ML and Robotics to transfigure agricultural policies by giving scalable, cost-efficient, and intelligent pest control solutions for sustainable agriculture.</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

Deploying YOLOv8m and Deep Q-Network for Intelligent Prediction of Pests and Soil-Borne Illness in Sugarcane Fields

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月3日

查看全文
  • M. Abinaya,
  • J. Alfred Daniel

摘要

In modern agriculture, integrating AI-oriented pest control and soil monitoring significantly improves crop productivity and sustainability. This article provides an innovative methodology for pest detection and control strategies and facilitates close-up soil monitoring with avoidance of soil-borne diseases in sugarcane fields by implementing integrated approaches of Deep Q-Network (DQN) and You Only Look Once Version 8 Medium (YOLOv8m) for object detection. The proposed work merges the features of YOLOv8m for effective pest identification with the existing dataset acquired from autonomous Soil Robots (ASR) designed for navigating and monitoring soil health conditions closely. YOLOv8m is used to recognize and categorize pests by giving a high precision rate in challenging environments. Here, drone-captured high-resolution pictures from the existing datasets of the sugarcane field have been used and processed by the proposed integrated models to acquire the best accuracy rate. This research mainly focuses on identifying major influencing pests like Early shoot borers, Black Bugs, Red Rust, Bacterial Blight, Stem Borers, Whiteflies, Red Rot, Leaf Spot, Mite Insect, and Crown Mealy Bugs based on the existing datasets. While, ASR with different sensors acquired data has been used and processed by the DQN algorithm to monitor the soil health based on the parameters of soil like pH, moisture, nutrient content, and temperature to avoid soil-borne diseases in sugarcane fields. The existing robots’ self-directed navigation abilities were offered by Reinforcement Learning, which allows the soil robots to adapt to various field conditions and prioritize regions with poor soil health conditions and pest detection hotspots. Experimental outcomes illustrate that the proposed integrated approach significantly enhances the accuracy of pest detection compared to traditional techniques by 98% and also provides detailed and close-up soil health monitoring. This existing data acquisition using drones and sensors allows better decision-making about pest control policies and soil handling measures, contributing to enhanced sugarcane crop health and productivity by the proposed integrated models. The proposed algorithm shows the feasibility of AI/ML and Robotics to transfigure agricultural policies by giving scalable, cost-efficient, and intelligent pest control solutions for sustainable agriculture.