Scene-level context-aware emotion recognition deep network method

Abstract

The invention discloses a scene-level context-aware emotion recognition deep network method, and the method comprises the steps: obtaining a body part image set XB by reading a training sample set Xinbody label value and an original emotion label value; normalizing the Xin and the XB, and respectively sending the normalized Xin and XB to an upper-layer convolutional neural network and a lower-layer convolutional neural network; extracting an emotion feature TF and a context emotion feature TC; sending TF and TC to an upper adaptive layer and a lower adaptive layer to obtain fusion weights lambda F and lambda C respectively; fusing TF, TC, lambda F and lambda C to obtain an emotion fusion feature TA, linearly mapping the TA through a full connection layer to obtain initial prediction values of arousal and valence, measuring loss between the two initial prediction values and an original emotion annotation value, implementing convergence step by step, completing training, and obtaining anetwork model; and processing the test sample set and sending the processed test sample set to a network model to obtain a predicted label value of a test sample set Xtn . According to the method, the influence degree of features of different attributes on the emotion of the person is considered when the features are fused, and the prediction performance of the model is improved on the basis of enriching emotion recognition research based on the image.

Description

technical field [0001] The invention belongs to the technical field of pattern recognition, and in particular relates to a scene-level context-aware emotional recognition deep network method. Background technique [0002] Emotion is a necessary form of expression for people to express their feelings. In daily life, understanding and identifying a person's emotions from the actual scene they are in can help to perceive their mental state and predict behavior, and interact effectively. As early as the 1990s, the concept of affective computing was proposed by the MIT Media Laboratory, and then scientists began to work on converting complex human emotions into numerical information that can be recognized by computers, so as to better realize human-computer interaction. Computers tend to be more intelligent, which has become one of the key issues to be solved urgently in the era of artificial intelligence. [0003] Traditionally, emotion recognition tasks for static images are ...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a scene-level context-aware emotion recognition deep network method, which solves the problem of the narrow range of emotion analysis based on static images in the prior art, only for face images, and the method of directly splicing different attribute features Limitations of Emotion Recognition

Images