Cross-age face recognition method based on online learning

Abstract

The invention discloses a cross-age face recognition method based on online learning, and belongs to the technical field of calculation, dead reckoning or counting. The method is realized by cascadinga model structure of an age feature separation network with a face feature extraction network. The face feature extraction network is trained on an external data set, so that the model has the capability of performing feature extraction on a face; the multi-task training of identity recognition and age prediction is performed on the age feature separation part by using the cross-age face data setto obtain an age-invariant face feature vector with higher robustness; in actual scene use, parameters of the feature separation part can be finely adjusted according to a newly input sample, and online learning of the neural network is realized. According to the method, the cross-age face recognition capability is improved to a certain extent while the model with low parameter quantity is used,online learning becomes possible due to the feature separation structure, and the model has the self-adaptive capability to a new scene.

Description

technical field [0001] The invention discloses a cross-age face recognition method based on online learning, relates to face recognition technology, and belongs to the technical field of calculation, calculation and counting. Background technique [0002] Face recognition is one of the most widely used and significant areas of computer vision tasks. Face recognition, as one of the biometrics, has the advantages of low cost, long-distance recognition, and no contact, compared with fingerprints, iris and other biometrics. In recent years, with the continuous development of deep learning, the face recognition algorithm based on deep convolutional neural network (Deep Convolutional Neural Network, DCNN) has shown a very powerful recognition ability, and the accuracy rate on the standard face verification data set LFW has reached Close to 100%. However, because age changes have a significant impact on facial features, general face recognition algorithms are generally not robust...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to address the deficiencies of the above-mentioned background technology, and to provide a cross-age face recognition method based on online learning, which improves the ability of cross-age face recognition while using a model with a low amount of parameters. Realize the self-adaptation of new samples in new scenes, and solve the technical problem that the robustness of face recognition algorithms to age spans is generally not strong

Images