Pathological image classification method based on deep learning two-stage inference network

The invention discloses a pathological image classification method based on a deep learning two-stage inference network. The tissue pathology image classification method at least solves the three difficult problems in a traditional tissue pathology image classification method that the calculation complexity is high, the difference of different areas is not considered, and local-global information is not effectively and synchronously concerned. In order to solve the problems, the method comprises the following steps: firstly, carrying out feature extraction on a tissue pathology image by using a relative position coding Vision Transform model; then extracting global information by using a global attention module, thereby effectively extracting feature representation of the tissue pathology image; and if the key information is not fully identified, finally, a differentiated local image region is extracted through a differentiable module in a fine reasoning stage, so that screening with finer granularity is real