UAV-Assisted Tea Leaf Blight Detection and Path Optimization for Targeted Pesticide Spraying

Tea leaf blight (TLB) poses a critical threat to tea production, particularly in Assam, the largest tea-producing state in India. Traditional manual inspection methods for TLB detection are labor-intensive, time-consuming, and often inaccurate, especially across large-scale plantations with diverse environmental conditions. Although advances in unmanned aerial vehicle (UAV)-assisted remote sensing have enabled high-resolution monitoring of tea plantations but the integration of detection outputs into geolocation-based precision spraying strategies is still underexplored. To bridge this gap, this paper proposes a two-stage UAV-assisted framework for effective TLB management. In the first stage, YOLOv5 is employed to accurately detect TLB-infected regions from UAV imagery. In the second stage, a deep Q-network (DQN) is used to determine the shortest route and minimize travel distance while ensuring complete coverage of infected zones for precision pesticide spraying. The proposed system allows for disease-specific treatment and is adaptable to both autonomous UAV operations and manual intervention, making it cost-effective and suitable for large-scale tea plantations. During training, experimental results show overall precision and mean average precision (mAP) of 71.51% and 88.96%, respectively.