Abstract:In this research

a retina image automaticrecognition system based on Convolutional Neural Network (CNN) is proposed forthe disadvantages of the traditional retina image processing process which iscumbersome and poor in robustness. First

image preprocessing includes noise removal

numericalnormalization

and data volume amplification; then

a new neural network model

XNet

is designed.XNet inherits the advantages of LeNet and Inception networks. Thenetwork parameters are based on training. The samples were adjustedadaptively. Finally

the comparison of accuracy and number of iterations was performed for differentnetwork structures.The experimental results show that the XNet network structure is superior toLeNet and Inception

and the accuracy rate can reach 91%; and the necessity ofdata amplification is confirmed through experiments.