Analysis of DenseNet201 with SGD optimizer for diagnosis of multiple rice leaf diseases

Abstract

Rice is the primary food crop worldwide. It serves as a source of energy and basic nutrition for more than half of the world's population. In rice leaf diseases such as blast, blight, and tungro affect the yield quality and quantity of rice grains. To overcome the above issues, a classifier was developed using a CNN model with SGD and Adam optimizer. The CNN classifier model extract the important features from the rice leaf image dataset and categories which the type of disease is affected in the paddy crop. The performance of the classifier was evaluated using a rice-leaf image dataset obtained from the Kaggle respository. According to the evaluation results, DenseNet201 with the SGD optimizer provided a maximum accuracy of 95% when compared to the Adam optimizer with 93.33% accuracy.