Method and system for image recognition

Abstract

The present invention relates to a method and system for picture recognition. The method includes: after receiving the picture to be recognized, inputting the picture to be recognized into a recognition model generated by pre-training for recognition, and outputting the recognition result; The correlation between the recognition result and the picture category, determine the picture category corresponding to the output recognition result, prompt when the picture category is recognized as a violent picture or pornographic picture, and mark and record the violent picture or pornographic picture to determine the violence Source of graphic or pornographic images. The invention can effectively identify indecent pictures on the network, greatly improves the accuracy of indecent picture identification, and is convenient for some network supervision departments or safety supervision departments to perform work.

Description

technical field The present invention relates to the technical field of image processing, in particular to a method and system for image recognition. Background technique Indecent images, abbreviated as NSFW (Not Suitable / Safe For Work, images that are inappropriate / unsafe to browse in the workplace), indecent images mainly include violent images and pornographic images. At present, there are already many technical solutions for identifying indecent images. In the field of image processing technology, traditional methods are currently used to judge whether an image is indecent. Whether the picture is a violent picture or not, for another example, for a pornographic picture, it is judged whether it is a pornographic picture based on the characteristics of sensitive parts of the human body combined with the text, etc. This method of identifying indecent pictures usually has a low recognition accuracy and cannot be effective It can accurately identify some indecent pictures sp...

AI Technical Summary

Problems solved by technology

At present, there are already many technical solutions for identifying indecent images. In the field of image processing technology, traditional methods are currently used to judge whether an image is indecent. Whether the picture is a violent picture or not, for another example, for a pornographic picture, it is judged whether it is a pornographic picture based on the characteristics of sensitive parts of the human body combined with the text, etc. This method of identifying indecent pictures usually has a low recognition accuracy and cannot be effective Accurately identify some indecent pictures spread on the Internet, which brings inconvenience to the work of some network supervision departments or security supervision departments

Images