Image processing method and device based on deep learning, and computer equipment

The invention relates to an image processing method and device based on deep learning, computer equipment and a storage medium. The problems of inaccurate classification and unfairness caused by data deviation and insufficient model fairness are solved. Firstly, diversity and comprehensiveness target image data are obtained, and rich training and testing samples are provided for a model. Then, the data are input into a model containing a multilayer convolutional neural network structure, a fairness loss item is introduced into a loss function of the model, and consistency and fairness of results are ensured; in this way, the model not only pays attention to the overall accuracy, but also optimizes the fairness among different groups. And finally, acquiring image processing result data output by the improved model, and improving the reliability and fairness of the model in practical application. In a word, by introducing a fairness loss item, the problem of unfairness in classification caused by deviation in t