Image processing method and device, image classification method and device, medium and electronic equipment

Abstract

The invention relates to an image processing method and device, an image classification method and device, a medium and electronic equipment. The image processing method comprises the steps of receiving a to-be-processed image; according to the to-be-processed image and an image processing model, a noise image corresponding to the to-be-processed image is determined, the image processing model comprises an image feature extraction sub-model and a noise information sub-model, and the image feature extraction sub-model is used for obtaining feature vectors of the to-be-processed image; the noise information sub-model is used for determining a noise image of the to-be-processed image according to a feature vector of the to-be-processed image, the image processing model in the training process further comprises a classification sub-model, and the loss of the image processing model is determined based on the noise vector added by the noise information sub-model and classification probability distribution output by the classification sub-model; according to the to-be-processed image and the noise image, a target image corresponding to the to-be-processed image is generated, and the to-be-processed image and the target image are the same in corresponding classification.

Description

technical field [0001] The present disclosure relates to the field of image processing, and in particular, relates to an image processing method, an image classification method, a device, a medium, and an electronic device. Background technique [0002] Network data security and risk control are currently a topic of great concern to content platforms. Content platforms need to be able to quickly identify potential risks while defending against malicious attacks. With the development of machine learning and deep learning technology and the improvement of computing power, existing attack methods are becoming more difficult to identify. [0003] In the prior art, when the content platform automatically reviews images, it usually trains an image classification model based on machine learning, so as to classify the images in the content platform based on the image classification model, thereby identifying the illegal content or Sensitive, close to real-time operations such as su...

AI Technical Summary

Problems solved by technology

[0004] At present, there are many hackers / attack technologies that may attack the content review mechanism, which can attack the image classification model without changing the overall shape of the image, and make the image classification model The modified image cannot be identified, thereby bypassing the review mechanism of the content platform and posing a greater threat to the security of the content platform

Images