Attribute face image generate method, device, system and readable storage medium

Abstract

The invention discloses a method for generating a face image with attributes, including receiving face feature description text and partial occlusion image; the description text being inputted to a text encoder network to obtain a text encoding vector; text coding vector and partially occluded image being cascaded to generate semantic feature data; the semantic feature data being input into the face image generation model to restore the face feature, and the face image with attributes being obtained. Among them, the face image generation model is a depth learning network which is optimized according to the text data set and the training set image training. The image generation method can generate a face image which conforms to the attributes described in the text according to the detail description of the text. The invention also discloses a face image generation device with attributes, a system and a computer-readable storage medium, which have the beneficial effects.

Description

technical field [0001] The invention relates to the field of computer vision research, in particular to a method, device, system and a computer-readable storage medium for generating a face image with attributes. Background technique [0002] The problem of image generation has always been one of the main issues studied by scholars at home and abroad. In the field of deep learning research, there is a greater demand for diverse face image datasets. How to use generative image technology to expand the data set and generate qualified high-definition face images according to specific attribute conditions is one of the problems that need to be solved urgently. [0003] At present, face image generation according to specific attribute conditions is based on attribute tags for image generation, such as pre-setting some attribute tags, "male", "female", "blond hair", "whether to smile", "whether to open mouth ", etc., generate the corresponding face image according to the selected...

AI Technical Summary

Problems solved by technology

[0004] However, using attribute labels (such as 5-dimensional attribute binary label vector: 00100) as a conditional limit for image generation has limitations

That is, the attributes of the image can only be limited to the range of the tags provided, and the diversity of attributes is poor, and the details of the definition of attributes through attribute tags are poor, and the description method is not routine

Images