System and method for generating face images of various styles of masks

Abstract

The invention relates to a system and a method for generating face images of masks with various styles, and the system comprises a facial feature detection module, a mask binary segmentation module and a mask pattern coding module, and the generated face images of the masks are optimized step by step. Wherein the facial feature detection module is mainly used for ensuring the correctness of the generated mask face image; the mask binary segmentation module is mainly used for retaining facial features of a non-occlusion area in an input image; and the mask style coding module is mainly used for extracting mask style codes from the input mask face and further guiding the model to generate masks with various styles according to the style codes. The problems that the face image shielding position is random, the shielding area is not consistent with the mask area, the face contour cannot be well reserved, the face recognition is not suitable for wearing the mask, the application scene is narrow and the like in the current face recognition image generation system can be solved.

Description

technical field [0001] The invention belongs to the technical field of image recognition, in particular to the field of face mask image generation, and in particular relates to a mask face image generation system and method with various styles, in particular to a correct, real and diverse type of mask face image generation system and methods. Background technique [0002] The history of the development of convolutional neural networks is not only a history of rapid improvement in image processing, but also a history of continuous improvement in face recognition technology. Face recognition is the most reliable way of human identification. When approaching the camera, it collects images or video streams containing human faces, and performs identification based on human facial feature information. Mask face recognition technology is a novel identification method. Compared with traditional face recognition models, mask face recognition has obvious advantages: convenience, saf...

AI Technical Summary

Problems solved by technology

[0004] In view of this, in order to solve the problem that the occlusion position of the face image generated by the current face recognition scheme is random and the occlusion area does not match the area of ​​the mask, the facial contour cannot be well preserved, and it is not suitable for wearing a mask for face recognition, and the application scene is narrow To solve this problem, provide a mask face image generation system and method with various styles; through the design of a progressive training mechanism that integrates three modules of facial feature detection, mask binary segmentation and mask style coding, the ability to generate mask face images is realized Correctness, authenticity, and richness of mask styles to significantly improve the performance of mask face recognition

Images