Enhanced Soybean Leaf Disease Detection through Deep Learning and Sine Cosine-Optimized Random Forest Classifier

Soybean is an important crop which faces threats to its diseases like Asian soybean rust, Frogeye leaf spot and Anthracnose which causes a great loss of yield. Conventional ways of detecting a disease are based on manual detection, and therefore the diagnosis is delayed and economical value is lost. This study outlines an automated identification system of the soybean leaf disease through deep learning and machine learning, namely, the optimization of the Random Forest during the Sine Cosine Algorithm (SCA). Considering that diseases can be recognized with the help of the system, the previously discussed feature extraction methods have been used: the Histogram of Oriented Gradients (HOG) and Convolutional Neural Networks (CNN). The training and evaluation dataset employed is UCI Soybean dataset and MATLAB has been used to implement the model. The proposed system is also able to attain 91.43% classification accuracy, which is a viable solution to early disease identification and reduction of pesticide application. The strategy can enhance agricultural practice since it offers real-time disease diagnosis, which is better, and sustainable farming.