Automatic Disease Detection in the Rice Leaves Employing a Support Vector Machine

Abstract

Rice is a major agricultural crop around the world. Crop diseases, on the other hand, have the potential to reduce yield and quality greatly, posing a major danger to global food supplies. As a result, disease control is essential for rice production. Accurate and prompt disease diagnosis is critical to disease control success, which allows pesticide control measures to be implemented. The most common method for diagnosing rice leaf diseases is a manual decision-making based on disease appearance. There aren't enough skilled workers in the area, for such tasks to be completed on time. As a result, a more effective and convenient way of identifying rice leaf diseases is required. Therefore, this research creates an automatic diagnosis approach for rice leaf disease detection using deep learning. The proposed solution is built with deep learning techniques and a huge dataset containing 2,000 images of three types of rice diseases such as leaf blast, sheath blight, and brown spot, and healthy leaf. The proposed model's robustness is improved by using its real-world rice leaf datasets as well as publicly available online datasets. With an accuracy of 96.0%, the proposed deep-learning-based strategy proved successful in automatically diagnosing the three discriminative diseases of rice leaves. Furthermore, 99.25% of the time, the algorithm accurately detected a healthy rice leaf. The results demonstrate that the suggested deep learning model gives a highly effective technique for identifying rice leaf infections, and is capable of quickly and reliably identifying the most common rice diseases.