Model Deep Learning Untuk Deteksi Penyakit Pada Tanaman Paprika (Capsicum Annuum L.), Tomat (Solanum Lycopersicum) Dan Strawberry (Fragaria Sp.).

Abstract

Penyakit yang menyerang tanaman dapat memberikan dampak yang serius terhadap produktivitas tanaman jika tidak dikendalikan dengan tepat. Cara efektif mengendalikan penyakit serta meningkatkan hasil panen adalah melakukan identifikasi dini penyakit tanaman secara akurat. Melihat perkembangan tersebut, metode deep learning sangat berpotensi untuk dikembangkan di Indonesia. Penelitian ini bertujuan mengembangkan model deep learning untuk mendeteksi penyakit pada tanaman paprika (Capsicum Annuum L.), tomat (Solanum Lycopersicum) dan strawberry (Fragaria Sp.). Penelitian ini menggunakan deep learning dengan algoritma YOLOv7 yang terdiri dari tahapan: (1) pengumpulan dataset, (2) pengolahan dataset, (3) pelabelan dataset, (4) training dataset, (5) uji model hasil training, dan (6) implementasi model. Training deep learning terhadap 9813 dataset dengan 4 kelas parameter penyakit yaitu bercak daun, hangus daun, hawar daun, dan virus kuning keriting dilakukan dengan menggunakan algoritma YOLOv7 dan YOLOv7-X. Hasil pengujian model deep learning untuk deteksi penyakit dengan menggunakan algoritma YOLOv7 dan YOLOv7-X menunjukkan tingkat akurasi sebesar 98,90% dan 98,98%, presisi 97,83% dan 98,12%, dan nilai recall 98,07% dan 98,63%. Berdasarkan penelitian ini dapat disimpulkan bahwa model cukup stabil dalam mendeteksi penyakit dengan tepat sehingga model deep learning untuk deteksi penyakit pada tanaman paprika, tomat, dan strawberry dapat digunakan.

Diseases that affect plants can have a serious impact on crop productivity if not properly controlled. An effective way to control disease and increase crop yields is by conducting early identification of plant diseases accurately. Based on these problems, the deep learning method has the potential to be developed in Indonesia. The aim of this study is to develop a deep learning model for detecting diseases of pepper plants (Capsicum Annuum L.), tomatoes (Solanum Lycopersicum), and strawberries (Fragaria Sp.). This research consists of several activities such as (1) datasets collections, (2) datasets pre-processing, (3) datasets labeling, (4) training deep learning models, and (5) testing and validating model. The model was developed using a deep learning model with YOLOv7 and YOLOv7-X algorithms using 9813 datasets with 4 classes of disease parameters, namely bacterial spot, leaf scorch, leaf blight, and yellow leaf curl virus. The results of the deep learning model test for plant diseases detection using YOLOv7 and YOLOv7-X algorithms showed accuracy rates of 98.90% and 98.98%, precision of 97.83% and 98.12%, and recall values of 98.07% and 98.63%, respectively. These results indicate that the deep learning model was capable to detect bell pepper, tomato, and strawberry plant with high accuracy and precision.