Paddy Disease Classification Using Deep Learning

Abstract

Bangladesh is mostly an agricultural nation. A large portion of people is dependent on agriculture. But lack of proper knowledge of our farmers to classify the disease the quality and quantity of our paddy are declining.One of the biotic variables that limit paddy production the most is the illness.These illnesses can result in severe decreases in agricultural production and quality, which can result in substantial financial losses for farmers.A key aspect of protecting paddy crops is early disease diagnosis. Through our research, we are attempting to discover paddy infections. This study focuses on neural network methods for classifying rice diseases and image processing methods for enhancing image quality. This methodology includes image acquisition, preprocessing, segmentation, analysis, and classification of paddy diseases.It is very challenging to detect any disease by merely looking at the leaves.We suggested a procedure and Our system uses a state-of-the-art method called image processing. To do this, we use a transfer learning classification algorithm that is based on the CNN (Convolutional Neural Network) method. In This, we used VGG16, MobileNetV2, and EfficientNetB3 algorithms, Whereas MobileNetV2 attained 90% accuracy, VGC16 attained 92% and the EfficientNetB3 model performed better than the other two models, outperforming them with a maximum accuracy of 98%, indicating a successful result of this study.