Faba Bean Disease Detection Using Deep Learning Techniques

Faba bean (Vicia faba L.) is an important cash crop for animal and human consumption in many countries, especially Ethiopia due to its high protein content and high rate of production. It also improves soil fertility with the help of bacteria, which forms nodules on its roots. Ethiopia is the second largest faba bean producer; however, according to research, the yield is generally low. Among the constraints contributing to low productivity of the crop, disease is the most important. Even though there are countermeasures such as the use of chemical sprays, which is used to reduce the damage or eliminate the disease, an effective disease identification technique is still required. Chocolate spot, faba bean gal, and rust are the most common diseases of faba beans. To effectively detect the disease, in this paper, we propose an end-to-end CNN model for training before and after the dataset is preprocessed. The CNN end-to-end model is trained with raw images and achieves 92.1% training accuracy. Also, the model is trained with preprocessed data and it achieves a training accuracy of 98.14% for the end-to-end CNN classifier. The results show that the proposed approach is suitable for faba bean disease detection and can be employed to detect diseases of other crops.