Emotion detection method based on baby expression recognition and cry

Abstract

The invention discloses an emotion detection method based on baby expression recognition and cry. The method comprises the steps: step 1, capturing face image information of a baby, and collecting anaudio data signal of the baby; step 2, when the baby is in a crying state, executing step 5; step 3, processing the captured face image; step 4, training the facial expression image obtained in the step 3 through a convolutional neural network to obtain facial expression features; step 5, windowing the infant crying voice signal obtained in the step 1 to generate a spectrogram corresponding to theinfant crying voice signal; step 6, training a classifier through a multi-layer perceptron and a support vector machine; step 7, inputting the facial expression features obtained in the step 4 into aclassifier for classification; and step 8, training an emotion detection model, and inputting the spectrogram in the step 5 into the trained emotion detection model to determine the emotion corresponding to the baby cry signal. According to the invention, the accuracy of infant emotion analysis is effectively improved.

Description

technical field [0001] The invention relates to the field of infant emotion detection, in particular to an emotion detection method based on infant expression recognition and crying. Background technique [0002] In 1976, Ekman and Friesen conducted in-depth research on microexpressions, (through observation and biofeedback) they mapped the correspondence between different facial muscle movements and different expressions to create FACS (Facial Movement Coding System). After research, Ekman believes that micro-expressions have a duration of no more than 1 / 5s and can reflect people's true emotions. These three characteristics are common in all human beings. [0003] Since the 1990s, after the optical flow method proposed by Mase and Pentland for facial expression recognition, the technology of automatic facial expression recognition has entered a new era. [0004] On March 31, 2014, an article published by researchers at Ohio State University in PNAS. The study used a compu...

AI Technical Summary

Problems solved by technology

[0006] Current technology often only uses facial recognition expressions of babies to analyze emotions, but ignores the complex emotions conveyed by babies when they cry

Emotional analysis using only a single facial expression cannot accurately express a baby's emotions

When the baby cries, the prediction effect is likely to change, and the stability of the prediction result cannot be guaranteed

Most of the existing technologies still analyze the baby's emotions through facial expressions, and few of them are combined with crying emotion detection to detect the baby's emotions.

Images