Detecting diseases in durian plants is a significant challenge in the agricultural sector, impacting yield and quality. This research aims to improve disease identification in durian plants by applying the Convolutional Neural Network (CNN) algorithm. The method involves collecting images of durian leaves, stems, and fruits infected by various diseases and using data augmentation techniques to expand the dataset and reduce overfitting. With the CNN model trained using this dataset, the accuracy reached 62% on validation data, with the highest accuracy for the “Healthy” class at 83%. The research results show that CNN is effective in recognizing diseases in durian plants, although there is still room for improvement through model optimization and the addition of training data. The implications of this research include the development of an AI-based disease detection system that can help farmers care for durian plants more efficiently and timely.
