Expression representation model training method, facial expression representation method and corresponding devices

Abstract

The invention provides an expression representation model training method, facial expression representation method and corresponding devices and relates to the technical field of image processing, and solves the technical problem of low accuracy of existing facial expression representation. The method comprises the steps that: a sample set is determined, and each sample in the sample set comprises a sample image and a sample label; and a to-be-trained expression representation model is trained by using the sample set, so that a trained expression representation model can be obtained, wherein the to-be-trained expression representation model includes a to-be-trained full face representation sub-model and a trained identity representation sub-model, wherein the trained expression representation model comprises a trained full face representation sub-model and a trained identity representation sub-model, and the output of the trained expression representation model is determined based on the difference value between the output of the trained full face representation sub-model and the output of the trained identity representation sub-model.

Description

technical field [0001] The present application relates to the technical field of image processing, in particular to a method for training an expression representation model, a method and a device for representing facial expressions. Background technique [0002] Humans have the instinct to perceive expressions, but this instinct is not possessed by machines. Accurate representations of expressions can promote the understanding of human emotions by machines, which provides a basis for building a friendly, intelligent, and harmonious human-computer interaction system. It is an important technical basis, and through the rational use of various expression representation methods, it can help promote the development of multiple related downstream tasks, including expression image retrieval, emotion recognition, facial action unit (Action Unit, AU) recognition, facial expression generation, etc. [0003] Currently, methods for detecting facial expression representations include AU-...

AI Technical Summary

Problems solved by technology

[0004] The purpose of this application is to provide a training method for an expression representation model, a facial expression representation method and a device, so as to alleviate the technical problem of low accuracy of the current facial expression representation method

Images