Method for detecting a facial micro-expression action unit based on a depth convolution neural network

Abstract

The invention discloses a method for detecting a facial micro-expression action unit based on a depth convolution neural network, which comprises the following steps: step 1, designing a depth convolution neural network structure; 1.1, marking that face and the rectangular shape area of different action units therein; 1.2, designing and realizing a depth convolution neural network, wherein that neural network comprise a convolution layer, a shortcut layer and an action unit detection layer, so as to learn the information of the face and the action unit area of different expression of the faceand the face, and obtain forward propagation parameters of the network; Step 1.3: taking the sample data in the face sample data set as the input data of the neural network; 2, detecting a facial expression action unit according to the network parameters learned in the step 1; Step 3: Visualize the output according to the face action unit detected in step 2. The detection method of the invention relies on a deep-layer convolution neural network to detect and recognize an action unit in a face image, and the detection accuracy and the detection speed can be improved.

Description

Technical field [0001] The present invention relates to the technical field of face recognition and emotion computing, in particular to a method for detecting human face micro-expression action units based on a deep convolutional neural network. Background technique [0002] Human face micro-expression is a natural expression of human inner emotions. Compared with ordinary expressions, micro-expression is not easy to be detected. It has the characteristics of small movement range and short stay time. The corners of the mouth are raised to express inner joy; The inadvertent curling up may hide the inner contempt; the upper lip is pushed up by the lower lip, which may hide a little dissatisfaction; the inner corner of the eye and the depression of the eyebrows may hide anger. These are the natural emotions of the micro expression Expression, because micro expressions are the leakage of real emotions, the detection and recognition of micro expressions is of great significance for cr...

AI Technical Summary

Problems solved by technology

However, the detection accuracy of action units using traditional geometric shape feature methods is very low, and action units are often affected by typical geometric structures such as the face, mouth, nose, eyebrows, etc., which are difficult to detect

Images