Implementasi Arsitektur YOLOv11 untuk Deteksi Penyakit Daun Tebu

Abstract

Sugarcane (Saccharum officinarum L.) plays an important role in the national sugar industry, but its productivity has declined due to leaf diseases such as mosaic, red rot, rust, and yellow leaf. Manual identification is often inefficient, especially for farmers in remote areas. This study proposes a YOLOv11 architecture for the detection and classification of sugarcane leaf diseases based on digital images, with performance analysis compared to previous deep learning models and the effect of image augmentation on accuracy. The dataset from the Sugarcane Leaf Disease Dataset on Kaggle includes 2,521 images with five classes (healthy, mosaic, red rot, rust, yellow). The data was processed through preprocessing, division (80% training, 10% validation, 10% testing), and augmentation (rotation, translation, flip). The results show an average precision of 97.2%, recall of 98.2%, mAP@0.5 of 98.8%, and mAP@0.5:0.95 of 95.5%, proving the effectiveness of YOLOv11 in accurate and fast detection.