Remote sensing image change detection method of convolutional neural network based on ResNet50 and DeeplabV3

The invention relates to a remote sensing image change detection method of a convolutional neural network based on ResNet50 and DeeplabV3 +, and solves the problems of low processing efficiency and poor processing precision of traditional remote sensing image change detection in the prior art. The remote sensing image change detection method of a convolutional neural network based on ResNet50 andDeeplabV3 + comprises the following steps: (1), performing data selection: selecting two remote sensing change detection images of a training set and a verification set for data selection; and (2), establishing semantic segmentation: establishing a semantic segmentation network based on the ResNet50 network by utilizing a deplabv3plusLayers function. According to the invention, the input layer ofthe network is modified; a large-range remote sensing image is directly processed by adopting a random patch and an image enhancement technology; the semantic segmentation convolutional neural networkis constructed by adopting