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

Character recognition model training method based on deep learning and recognition method thereof

A technology of deep learning and training methods, applied in character recognition, character and pattern recognition, instruments, etc., can solve problems such as inability to train, and achieve high accuracy

Active Publication Date: 2015-12-30
INST OF AUTOMATION CHINESE ACAD OF SCI +1
View PDF1 Cites 38 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, when the convolutional layer exceeds 5 layers, this method will lead to failure to train

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 recognition model training method based on deep learning and recognition method thereof
  • Character recognition model training method based on deep learning and recognition method thereof
  • Character recognition model training method based on deep learning and recognition method thereof

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0031] In order to describe the specific implementation of the present invention in detail, a word recognition data set is taken as an example. The data set contains 862 words in natural scenes that have been cropped, and each image contains a word and a small amount of background. The implemented model can automatically recognize words in images. Specific steps are as follows:

[0032] In step S1, 6113 character images are cut out from the word data set as a training set, and 5379 character images form a test set.

[0033] In step S2, a deep convolutional neural network with 5 convolutional layers + 3 fully connected layers is used for learning. The convolutional layer uniformly uses 128 nodes, a convolution window of 3×3, and a step size of 1. The number of nodes in the fully connected layer is 256, 256, and 62, respectively.

[0034] In step S3, the image training set is randomly divided into 8 subsets, and each subset contains 768 images (the last subset is less than 76...

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 provides a character recognition method based on deep learning. The character recognition method based on deep learning comprises the steps that a deeper multilayer convolution neural network structure is designed, and each character acts as a class; the convolution neural network is trained by using a back propagation algorithm so as to recognize the single character, the target function of the network is minimized in a supervised way and a character recognition model is obtained; and finally the most possible words are found out of the dictionary by adopting a viterbi algorithm according to the existing recognized characters. One input is given, candidate characters are obtained by performing sliding window scanning firstly and then the most possible words are found out of the candidate characters in testing. The character features are learned by utilizing the deeper convolution neural network, the method has robustness in character color, size, illumination and fuzziness, and relatively high accuracy of character recognition and word recognition can be maintained.

Description

technical field [0001] The present invention relates to the field of pattern recognition and machine learning, in particular to the technical field of neural network and deep learning, and more specifically to a text recognition model training method and recognition method based on deep learning. Background technique [0002] For text recognition (OCR), traditional methods require manually designed features, which require a lot of expert knowledge. Deep learning is a feature learning method. It only needs to provide a large number of training samples, and the model will automatically learn robust feature expressions. [0003] In addition, the common convolutional neural network convolution layer generally does not exceed 5 layers. Compared with a few convolutional layers, using more convolutional layers inevitably increases the difficulty of the problem. The traditional solution is to directly initialize the network weights randomly and then perform training. However, whe...

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): G06K9/00G06K9/62
CPCG06V30/10G06F2218/12G06F18/24
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 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