Class increment image classification method based on virtual feature generation and playback

The invention discloses a class increment image classification method based on virtual feature generation and playback, and aims to solve the problem that the classification accuracy of an existing image classification method needs to be improved. The method comprises the following steps: firstly, constructing a class increment image classification system based on virtual feature generation and replay, wherein the class increment image classification system is composed of a text prompt construction module, a text feature extraction module, an image feature extraction module, a virtual feature generation module, a virtual feature replay module and a main task module; and constructing a training set, a verification set and a test set. And training the image classification system by adopting the training set to obtain a trained image classification system. Verifying the trained image classification system by using the verification set, and assigning a weight parameter containing the best text prompt to the image