ANALISA PERFORMA CONVOLUTIONAL NEURAL NETWORK (CNN) DENGAN ARSITEKTUR YOLOV8 DALAM DETEKSI PENYAKIT DAUN TANAMAN HORTIKULTURA TROPIS

Tropical horticultural agriculture faces significant challenges due to leaf diseases that can drastically reduce crop yields. This study aims to evaluate the performance of a Convolutional Neural Network (CNN) algorithm using the YOLOv8 architecture for automatic leaf disease detection. The dataset consists of 1,000 images of tropical plant leaves (apple, banana, durian, mango, and orange), each categorized into two classes: healthy and unhealthy. The model was trained using a transfer learning approach with pretrained weights, and the training was conducted for 50 epochs. Evaluation was based on precision, recall, F1-score, and mean Average Precision (mAP) metrics. Results show that YOLOv8 demonstrates good detection performance, achieving a mAP of 0.92, and exhibits a stable convergence trend during training. The model is capable of identifying the location and category of leaf diseases in real time, providing high accuracy in object classification across various lighting conditions and backgrounds. This research confirms that YOLOv8 is a promising solution for computer vision-based plant disease detection systems to support the sustainability of tropical agriculture.