A Lightweight Network-Based Deep Binary Feature Facial Expression Recognition Method

Abstract

The present invention claims to protect a light-weight network-based deep binary feature facial expression recognition method, which belongs to the technical field of pattern recognition. The method mainly includes the following steps: first, constructing a set of convolutional neural network framework that binarizes parameters, and implanting the binary convolution mode into each residual network layer to form a bidirectional decision-making network model; then, Extract the LBP dynamic radius feature based on pixel gradient on the image input to the network, construct the LBP weight map with Huffman weight and the LBP binary map with Huffman weight; then use the LBP weight map, LBP binary map and the original image as the BRCNN network The multi-input features are constructed to construct deep binary features; finally, the deep binary features are cascaded for classification. The invention greatly reduces the amount of parameters during network training, reduces the computational cost of the network, enhances the expressive ability of features, and improves the robustness and speed of the method in facial expression recognition.

Description

technical field [0001] The invention belongs to the technical field of computer pattern recognition, in particular, to a facial expression recognition method. Background technique [0002] In the history of computer vision development, deep learning has become one of the most popular methods to solve computer vision problems; convolutional neural networks have also become one of the classic techniques of deep learning. From the emergence of LeNet and AlexNet networks to promote the development of deep learning, to deeper and wider convolutional neural networks such as VGGNet, ResNet, and InceptionNet, the extracted image features are developed in a more accurate and efficient direction. However, blindly deepening and widening the neural network cannot directly and effectively improve the accuracy and efficiency of visual tasks, but will increase the burden of equipment operation. How to design a "fast and good" lightweight network has become the focus of current research on...

AI Technical Summary

Problems solved by technology

However, how to achieve the stage of fast training and high recognition accuracy in the facial expression recognition task combining traditional feature operator methods and deep network features is still a difficult point

Images