Potato Leaf Disease Detection Using YOLOv8n with a Handheld Device

---

Abstract

This study addresses the urgent need for timely detection of leaf diseases in agriculture to mitigate potential crop losses and subsequent negative impacts on the global economy. We propose a deep learning-based technique for potato leaf disease recognition, utilizing the Ultralytics YOLOv8n model trained on PlantVillage datasets, with the aim of enabling detection using handheld devices like mobile phones or Internet of Things (IoT) devices. Evaluation metrics including precision, recall, F1-score, mean Average Precision (mAP) at different Intersection Over Union (IOU) thresholds demonstrate the robust performance of the model, with high values across all metrics, such as precision (94.57%), recall (94.164%), F1-score (97.297%), and mAP (94.367%). These outcomes underscore the efficacy of the YOLOv8n model in accurately finding diseased regions within potato leaves. The potential applications of this approach extend beyond disease detection alone, with implications for precision farming and crop management. Overall, this research contributes to advancing technologies aimed at improving agricultural productivity, resilience, and economic sustainability through efficient disease management strategies.

---

Keywords

Computer Vision, Deep learning, Image Classification, Object Detection, Plant leaf disease detection, Precision Agriculture, YOLOv8n.