Rice Disease Recognition Based on Improved Residual Network

The diagnosis of rice leaf diseases is of great significance to agricultural production and crop yield. With the rapid development of deep learning technology, deep learning has become an effective tool to solve crop disease diagnosis. In this paper, we propose a multi-scale convolutional neural network that combines improved CBAM attention mechanism with deep separable convolution to effectively diagnose common rice leaf diseases such as yellow dwarf, rice thrips, leaf char, brown spot, rice blast and bacterial leaf blight. The network structure combines CBAM module and ResNeXt50 residuals to accurately extract complex features of rice diseases. The experimental results show that the identification accuracy of the model on the rice leaf disease data set is as high as 99.66%, which is obviously better than the traditional model, and the diagnosis speed and accuracy are significantly improved. At the same time, the lightweight design of the model ensures a fast response and performs equally well on other publicly available crop leaf disease datasets, further verifying its good generalization ability and applicability. By adding CBAM module, the model can capture disease characteristics more efficiently and keep the parameter number low, providing a fast and efficient solution for rice disease diagnosis in actual agricultural environment.