SISTEM DETEKSI CITRA UNTUK IDENTIFIKASI PENYAKIT TANAMAN KELAPA SAWIT BERBASIS KECERDASAN BUATAN

Abstract

Palm oil is one of the main agricultural commodities in Indonesia. Oil palms are inseparable from disease attacks which can cause their growth to decrease. Limited information and knowledge of farmers hinders disease control. This research focuses on the need for a system that can help identify palm oil diseases. This research begins with collecting a dataset, then the data preprocessing stage is carried out. Next, the AI ​​model is developed, then the model is trained, validated and evaluated. After that, the model is implemented and tested and documentation and reports are created. During the 20 epochs carried out, there were variations in performance improvements. Although initially the model accuracy was still low, over time the model accuracy began to increase. The test results show that the process of detecting types of oil palm disease is very dependent on image quality, shooting distance, lighting conditions and noise in the image. Test results on a dataset consisting of 6 classes show an accuracy level of 98.63 percent. This image detection system provides good test results for recognizing new data. Thus, this system has proven to be effective and can be used to detect oil palm plant diseases.

Keywords: Artificial Intelligence, Image Detection, Classification of Oil Palm Diseases, Object Detection