Face recognition method and device based on residual quantization convolutional neural network

Abstract

In order to provide a face recognition method and device that can not only complete large-scale face recognition, but also reduce the amount of calculation to reduce hardware requirements and reduce the time required for training, the present invention provides a method based on residual quantized convolution The face recognition method of the neural network comprises the following steps: step S1, constructing a convolutional neural network model and training; step S2, preprocessing the target image and preprocessing the image to be determined; step S3, preprocessing the image to be determined The image and the preprocessed target image are sequentially input into the feature extraction model to obtain the feature vector to be determined and the target feature vector; step S4, to determine consistent face images according to the target feature vector and the vector to be determined, wherein step S1 includes setting the predetermined layer A step of approximating the parameter matrix of the quantization layer by performing integer bit quantization on the quantization layer parameters. The invention also provides a face recognition device based on residual quantization convolutional neural network.

Description

Technical field [0001] The present invention belongs to the field of machine learning, involving a face recognition method and apparatus, in particular to a face recognition method and apparatus based on residual quantization convolutional neural network. Background technique [0002] The human face recognizes the biometric information of the human face, and the identity of the corresponding person is identified by a certain technique, for example, it is determined from a large number of face images to be determined after obtaining the target character. Face image. Face recognition is an important research work in computer vision and mode identification, as well as a rich practical application scenario, such as community security, criminal pursuit, mobile payment, etc. [0003] The development of face recognition technology has been made for decades. Many related machine learning algorithms in the earlier year have been proposed, including methods based on geometric characteristi...

AI Technical Summary

Problems solved by technology

When applied to face recognition, for the 3-channel input of the size of 224×224 corresponding to the usual face image size, the calculation process includes more than 4.2 billion floating-point multiplications, so the storage and calculation overhead is very huge. When the application In large-scale face recognition (for example, when the number of images to be determined is huge), it is more likely to have problems such as excessive parameters, large models, massive calculations, and low efficiency. This not only increases the hardware requirements for face recognition, It will also increase the time required for model training and even make the training process almost impossible to complete, making it difficult for the convolutional neural network based on the residual learning mechanism to be practically applied to face recognition tasks

Images