au feature recognition method, device and storage medium

Abstract

The invention discloses an AU feature recognition method, device and storage medium. The method includes: taking a real-time image taken by a camera device, and using a face recognition algorithm to extract a real-time face image from the real-time image; The face image is input to a pre-trained facial average model, and the facial average model is used to identify t facial feature points from the real-time facial image; according to the positions of the t facial feature points, determine the relationship between each A feature area matched by an AU, extract local features from the feature area, and generate multiple feature vectors; input the multiple feature vectors into a pre-trained AU classifier that matches the feature area, and obtain from the The prediction results of the corresponding AU features identified in the feature region. Different AU classifiers are used to identify the AU features in the feature area of ​​the real-time facial image, which can effectively improve the efficiency of AU feature recognition.

Description

technical field [0001] The present invention relates to the technical field of computer vision processing, in particular to an AU feature recognition method, device and computer-readable storage medium. Background technique [0002] Facial emotion recognition is an important part of human-computer interaction and affective computing research, involving psychology, sociology, anthropology, life science, cognitive science, computer science and other research fields. significance. [0003] The internationally renowned psychologist Paul Ekman and his research partner W.V.Friesen have made in-depth research, and through observation and biofeedback, they have described the corresponding relationship between different facial muscle movements and different expressions. FACS is the "facial expression coding system" created in 1976 after years of research. According to the anatomical characteristics of the human face, it can be divided into several independent and interrelated motor ...

AI Technical Summary

Problems solved by technology

[0004] At present, the method of judging facial expressions by identifying AU features in facial images is relatively common and has a high accuracy rate. However, most of the industry’s identification of AU features is to collect a large number of AU samples, organize the samples, divide them into several categories, and use convolutional neural network The AU feature recognition model is trained by the network for AU feature recognition, but the accuracy of this method is not high

Images