Diabetic retinopathy (Diabetic Retinopathy, DR) is the symptom of diabetes affecting the retina during the onset of diabetes. Aiming at the problem of information loss of lesion areas such as microimages during model downsampling, this paper proposes a module of DenseNet fusion residual structure. This module first connects two consecutive Dense block, then sums the feature information using the residual structure, and processes the feature image information in parallel to prevent the loss of effective feature information. Finally, finally, the residual connects the two convolution blocks containing drop out to suppress the overfitting phenomenon. To solve the problem of the channel weighting of the feature graphs of lesion areas in previous convolution operations, the paper proposes a module of SeNet fusion residue structure. This module first connects SeNet, adds the feature information of global average pooling and global maximum pooling to improve the utilization of effective channel information, and then ensures the integrity of feature graph information through the residual mode of conv1x1. Based on the design of the above two modules, the paper proposes a DR classification method of DenseNet and SeNet fusion residue structures. The model achieves 89.8%precision and 97.0% specificity on the APTOS2019 dataset, 78.8% accuracy and 91.9% specificity on the Messidor-2dataset, which can effectively improve the classification ability of the degree of retinal lesions. [ABSTRACT FROM AUTHOR]