YOLOv8-Based Pest Detection for a Plant Treatment Spraying Machine in Agriculture

Abstract

Whitefly infestations pose a significant threat to chili crop productivity, leading to reduced yield and increased pesticide dependence. The aim of project is to develop an automated pest spraying machine that can effectively control whiteflies using advanced sensors and camera-based detection systems. By using advanced sensors and cameras, the system applies pesticides precisely, minimizing labor, chemicals, and environmental impact. A functional prototype detects and eliminates whiteflies, supported by a YOLOv8-based machine learning monitoring system. The system uses 3,150 images with computer vision for fast, accurate whitefly and anthracnose detection. The performance matrices for the whiteflies and anthracnose dataset are divided into three subsets: precision, mAP, and recall. This study achieves 75.27% precision, 64.3% mAP, and 69.84% recall. The results demonstrate the effectiveness of the YOLOv8 (You Only Look Once version 8) approach.  The results demonstrate that the YOLOv8 model ensures efficient, reliable pest detection and promotes sustainable, precise chili crop management.