Efficient neural network architecture method and system based on ERetinaNet

The invention discloses an ERetinaNet-based efficient neural network architecture method and system, and the method comprises the steps: selecting FRepVGG as a backbone network of a convolutional neural network model, and enabling the backbone network FRepVGG to be composed of a plurality of FRepVGG stages; equivalently converting a multi-branch structure of the FRepVGG block during training into a single-path structure during reasoning by using a structure reparameterization technology; feature fusion is carried out on the features of the middle layer in the FRepVGG stage; an effective multispectral channel attention module is introduced into the last layer of the FRepVGG stage; the method comprises the following steps of: inserting a Vision Transform module behind a backbone network; and a detection head of the original RetinaNet is properly simplified. The ERetinaNet convolutional neural network formed by the architecture provided by the invention is higher in breast mass detection efficiency and better in