Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Face recognition method based on convolutional neural network

A convolutional neural network, face recognition technology, applied in character and pattern recognition, instruments, computer parts and other directions, can solve the problems of poor self-adaptation, low accuracy, etc., to achieve enhanced robustness, short time cost, The effect of reducing complexity

Inactive Publication Date: 2018-05-25
YANSHAN UNIV
View PDF8 Cites 19 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to propose a face recognition method based on convolutional neural network in order to solve the problems of low accuracy, poor self-adaptation and manual parameter adjustment in existing face recognition methods for small data sets

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Face recognition method based on convolutional neural network
  • Face recognition method based on convolutional neural network
  • Face recognition method based on convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0021] Specific implementation mode one: as figure 1 As shown, a face recognition method based on convolutional neural network includes the following steps:

[0022] Step 1: Divide the face images to be recognized into three categories: training samples, test samples and verification samples, and read the data to be trained, each of which has its corresponding label;

[0023] Step 2: Normalize the face image after reading the training data in step 1;

[0024] Normalize the read face image, the reason is that the neural network is trained (probability calculation) and predicted by the statistical probability of the sample in the event, and the normalization is between 0-1 Statistical probability distribution, when the input signals of all samples are positive, the weights connected to the neurons of the first hidden layer can only increase or decrease at the same time, resulting in a very slow learning speed. In order to avoid this situation and speed up the network learning ...

specific Embodiment approach 2

[0029] Specific embodiment two: the difference between this embodiment and specific embodiment one is: the structure of the convolutional neural network is input layer-convolution layer-convolution layer-pooling layer-convolution layer-convolution layer-pool Layer-full connection layer arrangement, because the gradient of the ReLu function is not saturated and the calculation speed is fast, so the function converges faster, so we choose the ReLu function as the activation function, and the pooling method uses the maximum pooling method, such as figure 2 As shown, the output of the current layer is expressed as:

[0030] x e =f(u e )

[0031] u e =W e x e-1 +b e

[0032] where x e Indicates the output of the current layer, u e Represents the input of the activation function (the result after calculating the weight and bias of the current layer), f() represents the activation function, W e is the weight of the current layer, b e can be biased.

[0033] Other steps a...

specific Embodiment approach 3

[0034] Embodiment 3: The difference between this embodiment and Embodiment 1 or 2 is that in Step 3, Dropout is added after the pooling layer to disconnect the network, and the neurons are disconnected with a probability of 0.25 and 0.5. Connection.

[0035] Such as image 3 with Figure 4 As shown, the Dropout layer can randomly disconnect the connection of the network, which can effectively suppress the phenomenon of overfitting. Since the number of disconnected connections should not be too many, too few will also affect the effect, so we choose to disconnect the connections between neurons with the probability of 0.25 and 0.5.

[0036] Other steps and parameters are the same as those in Embodiment 1 or Embodiment 2.

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

No PUM Login to View More

Abstract

The invention relates to a face recognition method based on the convolutional neural network. The problems that an existing face recognition method for a small data set is low in accuracy and poor inself-adaptation and needs manual parameter regulation are solved. The method comprises the steps of 1, dividing face images needing to be recognized into training samples, testing samples and verification samples, and reading data to be trained; 2, performing normalization processing; 3, constructing a network structure of the convolutional neural network; 4, placing the training samples into thenetwork structure constructed in the step 3 for training, wherein the training process includes forward broadcasting and reverse broadcasting; 5, saving model parameters trained in the step 4; 6, adopting the trained model parameters for detecting the testing samples obtained after normalization processing in the step 2, and obtaining a model detection result. The method is used for the field of pattern recognition.

Description

technical field [0001] The invention relates to the field of pattern recognition, in particular to a face recognition method based on a small data set of a convolutional neural network. Background technique [0002] Face recognition has always been a hot spot in the direction of pattern recognition. There are four main face recognition methods: methods based on geometric features, methods based on models, methods based on statistics and methods based on neural networks. Convolutional neural network (CNN) is a neural network containing convolutional layers. In the 1960s, Hubel and Wiesel discovered its special features when studying the neurons used for local sensitivity and direction selection in the cat's cerebral cortex. The network structure can effectively reduce the complexity of the feedback neural network. Inspired by research, the model imitates the process of interactive processing of visual information between simple cells and complex cells in the visual cortex. S...

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
no application Login to View More
IPC IPC(8): G06K9/00G06K9/62
CPCG06V40/172G06F18/214
Inventor 黎捷宋建宏张梦达季淑梅
Owner YANSHAN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com