Rice leaf disease classification using depthwise separable convolutional neural network

Given the significant impact of rice leaf diseases on agricultural productivity, early and accurate disease identification is imperative. This study proposes the use of Depthwise Separable Convolutional Neural Networks (DSCNN) for efficient classification of rice leaf diseases. By effectively separating image pixels into different classes based on their characteristics, DSCNN provides a robust solution to this task. The proposed methodology involves preprocessing steps such as resizing, noise reduction using Laplacian, contrast enhancement using CLAHE, and data augmentation. The proposed CNN model has been evaluated with different loss functions along with varying learning rates. 20 iterations using RMSprop Optimizer with a learning rate of 0.0001 gave the best average validation and average test accuracy of 0.8659 and 0.9898 for our proposed CNN model. Based on the performance of our model, it is clear that the proposed model is capable of accurately diagnosing various rice plant diseases. Considering the small parameter size of 2.4 million, we conclude that the proposed model is a substantial improvement over traditional convolutional neural network architectures in rice plant disease detection.