Emoji package generation system and method based depth learning

Abstract

The invention relates to an emoji package generation system based depth learning. The system comprises a neural network control unit, a video processing module, a power module, a neural network training module, a neural network feedback module, a neural network test module, a test result post-processing module, a camera information processing module, a camera unit, a model storage module and a video database. The generative adversarial network (GAN) architecture is perfectly built, training difficulty is relatively low, the adversarial training method is employed by the GAN, the network does not directly replicate or average real data, so diversity of generated samples is enhanced, intelligent expression of a target portrait is realized, and generation of the above emoji package is realized in real time through automated calculation design.

Description

technical field [0001] The invention belongs to the technical field of emoticon package generation systems, in particular to a system and method for generating emoticon packs based on deep learning. The present invention utilizes face detection and face feature point marking technology to locate expressions, and uses depth Learning technology, the basis of deep learning is different from traditional image processing methods. It solves problems through computer autonomous learning, and can automatically build models according to problems. Background technique [0002] With the popularization of smart phones and the diversification of social software forms, in order to enrich the chat content and enhance the communication between users, many instant messaging software provides emoticon information, for example, many emoticon packages; users can choose various emoticons in the emoticon packages. Send information through emoticons, and express emotions and information that are d...

AI Technical Summary

Problems solved by technology

[0008] In order to solve the above-mentioned technical problems, the present invention provides a system and method for generating emoticons based on deep learning, which solves the problem that the existing portrait emoticons are recording and intercepting the expressions of a single user, and lack interest and interactivity. The generative confrontation network in the learning realizes the transplantation of expressions, allowing the user's expressions to be transplanted to another face to generate expression packs. The system of this invention can be based on deep learning, and according to the position of the original feature points, the entire human face can be generated. Face image, which really realizes the transplantation of expressions, which is more in line with real pictures, and can replace expressions in real time. The design is optimized and stable, and the work is stable;

Images