Convolutional neural network training method and apparatus for face recognition

Abstract

The invention discloses a convolutional neural network training method and apparatus for face recognition and belongs to the face recognition field. The method includes the following steps that: a sample training library is constructed; the sample training library is adopted to train a convolutional neural network; the trained convolutional neural network is adopted to extract the feature vectors of all samples in the sample training library; distances between every two feature vectors are calculated; a sample pair training library is constructed; the sample pair training library is composed of sample pair sets of all face image samples, wherein the sample pair sets include heterogeneous sample pairs and homogeneous sample pairs, wherein the heterogeneous sample pairs are composed of face image samples and face image samples which are separated from the face image samples by distances which are smaller than a certain value, and the homogeneous sample pairs are composed of face image samples and face image samples which are separated from the face image samples by distances which are larger than a certain value; and the sample pair training library is adopted to train a convolutional neural network. With the method adopted, recognition errors caused by make-up and external environment can be effectively avoided.

Description

Technical field [0001] The field of face recognition of the present invention particularly refers to a method and device for training a convolutional neural network for face recognition. Background technique [0002] With the rise of deep learning, especially the deepening of deep convolutional neural network research, a large number of network models based on Convolutional Neural Network (CNN) have been applied to image processing and image recognition, especially in face recognition The field has achieved remarkable results. [0003] In the field of face recognition and authentication, there are usually such problems. For example, due to makeup and external environmental influences, it may appear that the photos of two different people are very similar, and the two photos of the same person are quite different. Such abnormal samples are an important cause of identification errors. Summary of the invention [0004] The invention provides a training method and device for a convolu...

AI Technical Summary

Problems solved by technology

[0003] In the field of face recognition and authentication, there are usually such problems. For example, due to makeup and external environmental influences, two photos of different people may appear very similar, and two photos of the same person may be quite different.

Images