Classification of Crop Disasters Based on DenseNet

In agricultural production, crops are vulnerable to various diseases, resulting in yield reduction. Therefore, it is important to judge the condition of crops and deal with it timely in agricultural production. To address these issues, we use convolution neural network to establish a model for rapid classification of crop diseases. Based on the fine-tuning and migration learning training strategies with different weights, the DenseNet-201 network structure is used to train the data sets of collected crop disease images. The SE (Squeeze-and-Excitation Networks) module is added to the DenseNet network structure to solve the loss caused by the different importance of different channels of the feature map in the convolution pooling process, and the Leak ReLu function in the DenseNet network is used to improve the model fitting ability.The experimental results show that even for the complex data sets, the model can still achieve good results, and have a high accuracy of the target crop diseases.