Emotion recognition system and method for video images

Abstract

The present invention relates to the field of emotion computing, and proposes an emotion recognition system for video images, aiming to solve the problem of low accuracy of emotion recognition due to the loss of information between pictures. The system includes: a face extraction module, and an emotion sequence coding module , the emotional information modeling module, the emotional space mapping module, and the face extraction module divide the video image into frames, extract the face pictures from the framed pictures, and form a picture sequence; the emotional time-series coding module uses the 3D convolution algorithm to Emotional information calculation is performed on the front and rear frame information of each face picture in the sequence, and the temporal information code of emotion is generated; the emotional information modeling module receives the temporal information code, and based on the long-term dynamic characteristics of the emotional information, uses the convolution long-short-term memory model to determine the temporal information The emotional feature matrix; the emotional space mapping module receives the emotional feature matrix, maps it to the emotional space through full connection, and outputs the result of emotional recognition. Emotion recognition of video images is realized.

Description

technical field [0001] The invention relates to the field of emotion computing, in particular to the field of human face emotion computing, in particular to an emotion recognition system and method for video images. Background technique [0002] Affective computing is by giving computers the ability to recognize, understand, express and adapt to human emotions. The "emotion recognition" of the computer is to enable the computer to infer the emotional state by observing the premise environment of expression, behavior and emotion, so that the computer can communicate with people emotionally, so as to construct a friendly human-computer interface. The emotion recognition of the computer can carry out emotion recognition through features such as voice, facial features or language of the human face. For the description of emotions, emotions can be divided into discrete descriptions such as "happy", "sad" and "neutral"; and emotions can be mapped to continuous spaces, such as two...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above problems in the prior art, that is, in order to solve the problem of low accuracy of emotion recognition due to the loss of context-related information between pictures in the emotion recognition of video images, this application proposes a method for video Image emotion recognition system to solve the above problems:

Images