Real-Time Plant Disease Detection with YOLOv8n: A Lightweight Object Detection Approach

Khairy Hassan, Eslam Mohammed; Nefea, Sherif; Refaat, Hosam; Awad, Ahmed; Badry, Emad

doi:10.21608/sceee.2025.390772.1080

Real-Time Plant Disease Detection with YOLOv8n: A Lightweight Object Detection Approach

Document Type : Original Article

Authors

¹ Department of Electrical Engineering Suez Canal University,Egypt.

² Information System Dept., Faculty of Computers and Information- Suez Canal University, Egypt

³ Information System Dept., Cairo Higher Institute for Languages and Simultaneous Interpretation, and Administrative Science, Egypt

10.21608/sceee.2025.390772.1080

Abstract

Plant diseases pose a significant risk to agriculture worldwide, particularly in areas that heavily rely on rice production, leading to yield losses and financial setbacks. This research presents a real-time rice disease identification system based on YOLOv8n, a rapid and resource-efficient object detection model. The model was trained on a manually labeled rice disease dataset from Kaggle and demonstrated strong performance, achieving a mAP@0.5 of 91.2%, a mAP@0.5:0.95 of 63.7%, and an average inference time of 15.6 milliseconds per image when run on a GPU.For practical field deployment, the model was implemented on a Raspberry Pi 4 integrated with a camera and touchscreen interface. This portable, low-cost system enables real-time, offline disease detection, making it especially suitable for use in rural and underserved regions. By delivering immediate and reliable diagnostics, the system enhances early response strategies and contributes to more resilient and sustainable agricultural practices.
Plant diseases pose a significant risk to agriculture worldwide, particularly in areas that heavily rely on rice production, leading to yield losses and financial setbacks. This research presents a real-time rice disease identification system based on YOLOv8n, a rapid and resource-efficient object detection model.

Keywords

Main Subjects