Cancer tissue pathology image classification method and system, medium, equipment and terminal

The invention belongs to the technical field of image classification, and discloses a cancer tissue pathology image classification method and system, a medium, equipment and a terminal, and the method comprises the steps: extracting features through employing a self-supervision pre-training model based on a target data set based on the self-supervision pre-training of a constructed positive sample pair strategy; a full-section pathological image sample feature matrix is constructed by using sample features with high attention scores, and feature mixing enhancement multi-instance learning based on attention is realized. The features extracted by using the self-supervised pre-training model based on the target data set are more suitable for the field of histopathological images, and the features learned by the model are more representative. According to the method for constructing the positive sample through comparative learning, the sample diversity is improved, and the model can learn better feature represent