Face recognition method and device
A technology of face recognition and face image, which is applied in the field of artificial intelligence, can solve problems such as the decrease of calculation accuracy, affect the accuracy of face recognition, and high calculation intensity, so as to improve the speed, avoid the decrease of accuracy calculation, and ensure high efficiency and accuracy effects
Active Publication Date: 2021-10-29
第六镜科技(北京)集团有限责任公司
View PDF13 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0002] The convolutional neural network is widely used in face recognition, and the calculation of the convolutional layer in the forward propagation of the convolutional neural network will account for 90% of the calculation of the entire network. Generally, the high-precision convolutional neural network contains The large number of parameters leads to relatively high computational intensity, which cannot meet the needs of face recognition application scenarios that require high real-time response
[0003] Although the existing acceleration methods for convolutional neural networks can increase the computing speed of convolutional neural networks and thus improve the efficiency of face recognition, the calculation accuracy will decrease due to value overflow during the accelerated computing process, and eventually Affected the accuracy of face recognition
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
specific Embodiment approach
[0087] As a specific implementation, the feature extraction network includes a quantization layer, a convolutional layer, and an inverse quantization layer; when the feature extraction module 120 is used to extract features from an input feature map using the feature extraction network to obtain an output feature map, it is specifically used for : The quantization layer quantizes the input feature map according to the predetermined first parameter, and outputs the quantized feature map. The first parameter is determined according to the number of channels of the input feature map and the number of preset channel groups; the convolution layer according to the predetermined The quantized weight parameter performs convolution processing on the quantized feature map, and outputs an intermediate feature map, wherein the quantized weight parameter is obtained by quantizing the preset weight parameter according to the predetermined second parameter; the inverse quantization layer is ba...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The present invention relates to the field of artificial intelligence technology, and provides a face recognition method and device. The method includes: acquiring a face image to be recognized; inputting the face image into a pre-trained face recognition model, and using face recognition The model processes the face image to extract the face features of the face image, wherein the face recognition model includes at least one feature extraction network, and the feature extraction network sequentially performs quantization, convolution and dequantization on the input feature map, quantization and The first parameter used in dequantization is determined according to the number of channels of the input feature map and the number of preset channel groups; face recognition is performed on the face image according to the face features. The invention not only improves the speed of convolution operation, but also avoids the decrease of accuracy calculation caused by numerical overflow in the convolution operation process, and finally ensures the high efficiency and accuracy of face recognition.
Description
technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a face recognition method and device. Background technique [0002] The convolutional neural network is widely used in face recognition, and the calculation of the convolutional layer in the forward propagation of the convolutional neural network will account for 90% of the calculation of the entire network. Generally, the high-precision convolutional neural network contains The large number of parameters leads to relatively high computational intensity, which cannot meet the needs of face recognition application scenarios with high real-time response requirements. [0003] Although the existing acceleration methods for convolutional neural networks can increase the computing speed of convolutional neural networks and thus improve the efficiency of face recognition, the calculation accuracy will decrease due to value overflow during the accelerated computi...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V40/168G06V40/172G06V10/44G06N3/045G06F18/241
Inventor 张义夫刘闯叶雨桐胡峻毅陈诗昱
Owner 第六镜科技(北京)集团有限责任公司