Sugar net disease auxiliary diagnosis method and device with limited computing resources

The invention provides an auxiliary diagnosis method and an auxiliary diagnosis device for a computing resource-limited diabetic retinopathy, a residual neural network model is adopted to perform classification prediction on an eye fundus image of a patient, the residual neural network model is improved, residual blocks are improved, shallow layer features are extracted, and the model has very strong relation induction deviation, so that the calculation resource-limited diabetic retinopathy diagnosis method and the auxiliary diagnosis device have the advantages that the calculation resource-limited diabetic retinopathy diagnosis efficiency is improved, and the calculation resource-limited diabetic retinopathy diagnosis efficiency is improved. The improved model can learn more features, feature expression can be better carried out, the auxiliary diagnosis task of the diabetic retinopathy is better carried out, and the classification precision can be finally improved. Besides, the model is simple in structure,