Character detection method and device based on deep learning

A text detection and deep learning technology, applied in neural learning methods, character and pattern recognition, instruments, etc., can solve problems such as increasing computing time, and achieve good results

Active Publication Date: 2015-12-23
INST OF AUTOMATION CHINESE ACAD OF SCI +1
View PDF4 Cites 62 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] In addition, the traditional scanning window method needs to test each window, and there will be overlap between each window, which will greatly increase the calculation time

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
  • Character detection method and device based on deep learning
  • Character detection method and device based on deep learning
  • Character detection method and device based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0047] In order to describe the specific implementation of the present invention in detail, a character detection data set is taken as an example. The dataset contains 250 natural scene images containing text as a training set and 249 as a test set. The implemented model can automatically detect text inside images. Specific steps are as follows:

[0048] Step S1, cut out 5980 character images from the data set as a training set, and 5198 character images form a test set.

[0049] Step S2, use a 2-layer convolutional layer + 3-layer fully connected layer deep convolutional neural network for learning, the first convolutional layer uses 64 feature maps, a 9×9 convolutional window, and the second convolutional layer Using 64 feature maps, a 5×5 convolution window, and a convolution step size of 1. The number of nodes in the fully connected layer is 128, 128, and 62 respectively, such as figure 2 shown.

[0050] In step S3, the stochastic gradient descent method is adopted, ...

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 discloses a character detection method and device based on deep learning. The method comprises the steps: designing a multilayer convolution neural network structure, and enabling each character to serve as a class, thereby forming a multi-class classification problem; employing a counter propagation algorithm for the training of a convolution neural network, so as to recognize a single character; minimizing a target function of the network in a supervision manner, and obtaining a character recognition model; finally employing a front-end feature extracting layer for weight initialization, changing the node number of a last full-connection layer into two, enabling a network to become a two-class classification model, and employing character and non-character samples for training the network. Through the above steps, one character detection classifier can complete all operation. During testing, the full-connection layer is converted into a convolution layer. A given input image needs to be scanned through a multi-dimension sliding window, and a character probability graph is obtained. A final character region is obtained through non-maximum-value inhibition.

Description

technical field [0001] The invention relates to the technical field of pattern recognition and machine learning, in particular to a method and device for character detection based on deep learning. Background technique [0002] For text detection, traditional connected region methods such as SWT or MSER think that text is connected, and these methods do not handle blurred text well. Deep learning is a region-based method. We only need to provide a large number of training samples, and the model will automatically learn robust feature expressions, which can handle ambiguous situations well. [0003] In addition, the traditional scanning window method needs to test each window, and there will be overlap between each window, which will greatly increase the calculation time. Considering that the convolution operation has translation invariance, the fully connected layer can be converted into a convolutional layer, and the entire test image can be convolved to obtain the probabi...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/08
CPCG06N3/088G06F18/2413G06F18/214
Inventor 王亮王威张宇琪范伟
Owner INST OF AUTOMATION CHINESE ACAD OF SCI
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap