Improved neural prototype tree-based interpretable fine-grained image classification method and system

The invention belongs to the technical field of image processing, provides an interpretable fine-grained image classification method and system based on an improved neural prototype tree, and designs an interpretable fine-grained image classification model comprising a multi-granularity feature extraction layer, a prototype layer and a soft neural binary decision tree layer. Obtaining feature representation of a to-be-classified image through the multi-granularity feature extraction layer, and generating a depth feature map; the prototype layer calculates the similarity between the prototype and the patch according to the depth feature map, and finds out the patch closest to the prototype; replacing each prototype with a potential patch closest to the prototype, carrying out visual display, taking the potential patch as an evidence, training the prototypes by the soft neural binary decision tree layer, removing wrong prototypes by using a background prototype removal mechanism, making a prototype path decisio