Network social expression package synthesis method based on image-text semantics, electronic equipment and computer readable storage medium

Abstract

The invention discloses a network social expression package synthesis method based on image-text semanteme, electronic equipment and a computer readable storage medium, and the method comprises the following steps: 1, collecting an expression package picture, and carrying out the semantic annotation of the expression package picture; 2, converting the semantic annotation into a text description vector and a word feature vector; 3, processing the word feature vector to obtain an augmented vector; 4, training the DetailGANj network to obtain content blocks corresponding to the augmentation vectors; 5, performing down-sampling on each content block to obtain a joint vector, and processing the joint vector to obtain an augmented vector; and 6, training the JoinGAN network by using the augmented vector, and synthesizing an expression package picture. According to the method, the synthesis model can be trained by using fewer sample pictures, and the expression package picture synthesized by using the model is relatively high in semantic similarity with the text and is more accurate.

Description

technical field [0001] The invention belongs to the technical field of natural language processing and computer vision, and relates to a method for synthesizing network social emoticons based on graphic and text semantics, electronic equipment and a computer-readable storage medium. Background technique [0002] With the development of the Internet and the prevalence of online chat tools, a new culture has emerged—emoticon pack culture. Online emoticon packs meet the most instinctive expressions and emotional needs of human beings. The system update has a subtle influence and promotion, which has played a very positive role in social decompression. In recent years, Doutu has become more and more popular, the number of users has continued to increase, and the scale of the industry has continued to expand, attracting a large number of capital and talents to join . [0003] In recent years, developers have developed various map-making and map-fighting applications in response ...

AI Technical Summary

Problems solved by technology

[0004] Text synthesis emoticons are a major application of text-to-image synthesis, which is an interdisciplinary field of computer vision and natural language processing. The main goal of text synthesis emoticons is to generate high-resolution and semantically consistent images. Currently, there are The text-to-image synthesis framework can only generate higher-resolution images, but there is a lack of semantic consistency. The synthesized image has a low semantic similarity with the text, and the accuracy and fitting degree are poor. There are relatively large restrictions on development and promotion

Images