In recent years, deep learning has revolutionized machine learning and has been used with great success in various engineering fields, such as transportation, agriculture, finance, and marketing. The interest in deep learning for such applications is due to its ability to manage gigantic volumes of data and model complex interrelated systems in order to improve decision-making, reduce costs, and optimize resources. In the field of agriculture, plant diseases affect the growth of various species, hence there is a need to identify and treat them early. In this work, we propose an incremental learning method based on deep neural networks for the classification of plant diseases. The goal of incremental learning is to allow the learning model to adapt to the arrival of new data without forgetting its previously existing knowledge. Our method is based on preserving the diversity of old data, some of which will be used when training the network for new tasks. To test the effectiveness of the approach, we tested it on the PlantVillage Dataset. The experimental results have shown that this method is efficient and provides better results than the incremental model iCaRL.