Machine learning-based method and system for forming facial expression intensity calculation model

Abstract

The invention discloses a machine learning-based method and system for forming a facial expression intensity calculation model, and the method and system are used for estimating facial expression intensity under expression classification, and improving the accuracy of facial expression intensity calculation through feature fusion and integrated learning. The method comprises the following steps: firstly, preprocessing acquired facial image data in an expression database to obtain facial feature points corresponding to a facial part; then, according to the obtained coordinate positions of the facial feature points, cutting the facial expression area of each piece of facial image data to obtain a corresponding facial expression image; and finally, performing facial feature extraction on theextracted facial expression image, and constructing an expression intensity estimation model. When one piece of image data containing the facial part is input into the expression intensity estimationmodel, the relation between the facial features and the expression intensity can be calculated.

Description

technical field [0001] The present invention relates to the fields of image processing and expression recognition, and in particular, relates to a method and system for calculating expression intensity through facial information. Background technique [0002] With the development of image processing and machine learning, expression recognition technology has made great achievements. However, most of the current expression recognition research is still doing expression classification (such as happiness, sadness, doubt, fear, sadness, etc.), and seldom estimates the intensity of expressions (such as: happy, happy level; angry, angry level). Facial expression intensity estimation can better calculate people's emotions and obtain more accurate cognition of people's emotions. Contents of the invention [0003] The technical problem to be solved by the present invention is to provide a method and system for calculating expression intensity based on expression classification an...

AI Technical Summary

Problems solved by technology

However, most of the current expression recognition research is still doing expression classification (such as happiness, sadness, doubt, fear, sadness, etc.), and seldom estimates the intensity of expressions (such as: happy, happy level; angry, angry level)

Images