DETEKSI KERUSAKAN DAN KUALITAS BUAH KAKAO (COKLAT) SECARA OTOMATIS MENGGUNAKAN PEMANFAATAN CONVOLUTIONAL NEURAL NETWORK (CNN)

Abstrak

Tanaman kakao (Theobroma cacao L) merupakan komoditas penting di maliku utara, namun prodiktifitas dan kualitas buah sering menurun akibat serangan hama dan penyakit yang sulit dideteksi secara dini. Penilaian kualitas buah secara manual masih juga bersifat subjektif dan kurang efesien. Penelitian ini bertujuan merancang sistem deteksi kerusakan dan kualitas buah kakao secara otomatis menggunakan metode deep learning dan arsitektur convoluational neutal network (CNN). Dataset berupa citra buah kakao yang di ambil langsung di perkebunan desa lelilef woebulen dengan variasi pencahayaan alami dan di ambel di kagle dan di klasifikasikan kedalam lima kelas: buah sehat, buah mentah, buah busuk, buah rusak akibat hama, buah rusak akibat penyakit. Model CNN di latih menggunakan TensorFlow / keras dan diuji secara real time melalui webcam. Hasil pengujian menunjukan sistem mampu mengklasifikasikan kondisi buah kakao dengan akurasi yang cukup tinggi. Sistem ini diharapkan dapat membantu petani dalam deteksi dini kerusakan dan meningkatakan kualitas serta hasil produksi kakao.

Cocoa plant (Theobroma cacao L) is an important commodity in north maluku; however, its productivity and fruit quality often decline due to pest and disease attacks that are difficult to detect at an early stage. Manual assessment of fruit qualiti is also subjective and inefficient. This study aims to design an automatic detection system for cocoa fruit damage and quality using deep learning with a convolutional neural network (CNN) architecture. The dataset consists of cocoa fruit images captured directly in plantations in lelilef village under varying natural lighting conditions and classified into five categories: healthy fruit, unripe fruit, rotten fruit, fruit damaged by pests, and fruit damaged by diseases. The CNN model was trained using TensorFlow/ keras and tested in realntime using a webcam. The results show that the system is able to classify the to assist farmers in early detection of fruit damage and improve the quality and productivity of cocoa