Chinese text description face image generation method based on face features

Abstract

The invention discloses a Chinese text description face image generation method based on face features. The method is characterized by comprising the following steps of carrying out sentence segmentation, word segmentation, part-of-speech analysis, filtering and dictionary mapping on a Chinese text for describing the human face features; extracting the feature vectors from the sentences and words of the text, realizing the conversion from a Chinese text sequence to mathematical representation, combining the extracted feature representation with a face image template, inputting the extracted feature representation into a deep neural network model, and generating a face image matched with the text description. Compared with the prior art, the generation of the face image conforming to the descriptive features of the corresponding Chinese text has stronger pertinence; the part-of-speech filtering operation is applied during the process of processing the description text, the semantic filtering and the fusion are carried out through the automatic encoder, the interference of a task-independent and meaningless text in the description text is effectively eliminated, the method is simple and is easy to implement, and the randomness of the natural language description is better solved.

Description

technical field [0001] The invention relates to the technical field of natural language processing and image generation, in particular to a method for generating face images based on human face features described in Chinese text. Background technique [0002] In some specific scenarios, it is necessary to quickly obtain facial portraits of people based on given partial information, including the task of generating facial images based on descriptive text in natural language. Usually, this task is often completed and checked manually by humans due to the vague content and information described. However, with the development of machine learning methods, it is possible to input facial features of people into computers to obtain facial images. The input of earlier facial image methods is mostly some simple phrase descriptions or information in artificially specified formats, such as determining that the hair color is one of the given color lists such as black, red, and yellow, an...

AI Technical Summary

Problems solved by technology

[0003] The description of natural language in the prior art is quite arbitrary, and it is difficult to fully express it directly with simple attributes, and the current method is based on languages ​​such as English, which have important differences from Chinese grammar, word format, etc. , related methods are difficult to generalize, so it is urgent to develop related models to extract features and generate images.

Images