Facial expression recognition method based on cross-connection multi-feature fusion convolutional neural network

Abstract

The invention discloses a facial expression recognition method based on cross-connection multi-feature fusion convolutional neural network, comprising the following steps: first, performing face detection and data preprocessing to reduce background information interference; then, building a cross-connection multi-feature fusion The convolutional neural network automatically extracts rich and effective facial expression features, and fuses the high and low-level features of the network; finally, the softmax is used for facial expression classification. The present invention fuses high-level semantic features and low-level features of facial expression images, and makes full use of the feature information learned by each hidden layer to extract more adequate and detailed expression features, and solves the problem of poor expression recognition effect, The problem of poor robustness.

Description

technical field [0001] The invention belongs to the technical field of facial expression recognition based on deep learning, and in particular relates to a facial expression recognition method based on a cross-connection multi-feature fusion convolutional neural network. Background technique [0002] The general steps of facial expression recognition include face detection, image preprocessing, expression feature extraction and expression classification. The current algorithm research on face detection, image preprocessing and expression classification has been relatively mature, and the core link of expression feature extraction has become the current research focus. In the traditional facial expression recognition method, using hand-designed expression features, which relies on professional knowledge and luck, it is difficult to extract comprehensive information, and it is sensitive to lighting changes, etc., and is not robust enough in real scenes. As an algorithmic tool...

AI Technical Summary

Problems solved by technology

At the same time, the continuous deepening of the number of network layers will cause the problem of gradient dispersion, which will also have a certain impact on the training of the network.

Images