Paddy Disease Classification Based on the Lightweight MobileNet-V2

Abstract

Paddy, a critical global staple food and economic resource, faces significant challenges in cultivation due to diseases that have devastating effects on crop yield and quality. Accurate and valid classification of these diseases is essential for their effective prevention, management, and timely treatment. Traditional identification methods, which are manual, time-consuming, labor-intensive, and prone to misclassification, fail to efficiently address these concerns. In response to these limitations, this research focuses on designing a lightweight neural network model for rice disease classification, leveraging the power of MobileNet-V2. This deep separable convolution-based neural network architecture is optimized for efficiency and accuracy in image classification tasks, making it well-suited for mobile devices. The approach enables real-time identification of rice diseases in the field, facilitating prompt intervention and treatment, ultimately minimizing the impact on crop yield and quality. The proposed model has undergone rigorous testing and benchmarking against state-of-the-art methods in paddy disease classification. Sufficient and multifaceted results demonstrate that the designed method achieves superior performance, outperforming the state-of-the-art in terms of accuracy and efficiency. The utilization of MobileNet-V2 in this research offers a valuable solution for the rapid and precise diagnosis of rice diseases, significantly contributing to the minimization of their spread and impact on crop yield and quality.