Picture content auditing method and device, equipment and medium

Abstract

The invention discloses a picture content auditing method and device, equipment and a medium, and the method comprises the steps: obtaining a to-be-audited picture in response to a picture auditing request; calling a scene recognition model to perform scene recognition on the to-be-audited picture to obtain a scene category of the to-be-audited picture; and according to the scene category of the to-be-audited picture, calling a negative model corresponding to the scene category to perform content auditing to obtain an auditing result of the to-be-audited picture. The technical problem that the false alarm rate and the missed detection rate of the auditing result are high in the prior art is solved.

Description

technical field [0001] The present application relates to the technical field of image processing, and in particular to a method, device, device and medium for reviewing picture content. Background technique [0002] Image content review is to review whether the image contains illegal information, such as information related to pornography, explosives, terrorism, politics, gambling, and user privacy. In content review, there are often multiple violation review categories, such as political review, pornographic review, personal privacy information identification, etc., and each type of violation identification is difficult to meet business requirements in general scenarios. For example, due to blurred faces, side faces, comic faces, etc., political-related recognition often leads to missed inspections of political-related pictures; in pornographic review, it is often difficult to distinguish pornographic from sexy, resulting in missed reviews or misjudgments by machines; for ...

AI Technical Summary

Problems solved by technology

In content review, there are often multiple violation review categories, such as political review, pornographic review, personal privacy information identification, etc., and each type of violation identification is difficult to meet business requirements in general scenarios

For example, due to blurred faces, side faces, comic faces, etc., political-related recognition often leads to missed inspections of political-related pictures; in pornographic review, it is often difficult to distinguish pornographic from sexy, resulting in missed reviews or misjudgments by machines; for personal privacy Information recognition, handwritten user privacy information often exists in real scenes, and existing technologies are difficult to identify these handwritten content, resulting in the leakage of personal privacy information

Images