Expression recognition network training method and system combined with weak supervision, medium and terminal

Abstract

The invention provides an expression recognition network training method and system combined with weak supervision, a medium and a terminal. An expression recognition network comprises a feature map extraction network, a feature extraction sub-network, a feature map matching sub-network and a classification sub-network. The method comprises the following steps: training the feature map extractionnetwork; and training the feature map matching sub-network and the classification sub-network. According to the invention, weak supervised learning is carried out by introducing the facial expressionfeature map. The accuracy and robustness of facial expression recognition and classification are greatly improved, so that the invention can adapt to facial expression recognition of facial images invarious scenes such as different angles, distortion and occlusion, the purpose of multi-task learning of an expression recognition network is achieved, and the accuracy of facial expression recognition is improved; therefore, by recognizing the facial expression, customer satisfaction can be analyzed, and fatigue detection or psychological treatment can be carried out on a driver.

Description

technical field [0001] The invention belongs to the technical field of facial expression recognition, in particular to an expression recognition network training method combined with weak supervision, a system, a medium and a terminal. Background technique [0002] Facial expression is an important way to express emotions and communicate information. Facial Expression Recognition is of great significance in many human-computer interaction systems, such as social robots, driver fatigue detection, customer satisfaction detection, medical treatment, etc. In 1971, the research of psychologists Ekman and Friesen first proposed that human beings have six main emotions, and each emotion reflects a unique psychological activity of a person with a unique expression. These six emotions are called basic emotions, which are Anger, happiness, sadness, surprise, disgust, and fear, like other computer vision problems, facial expression recognition also faces many challenges, such as head ...

AI Technical Summary

Problems solved by technology

[0010] The CNN-based method has greatly improved the effect compared with the traditional method. It has a stronger description ability than artificial features and can effectively capture the features of human facial expressions. However, due to the small degree of differentiation between expressions, the existing The method's ability to distinguish facial expressions needs to be improved, so that accurate classification cannot be achieved

Images