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

Myancanda image text recognition method based on CRNN

A text recognition and image technology, applied in character recognition, neural learning methods, character and pattern recognition, etc., can solve the problem of difficult extraction of text information in Burmese images, achieve good results, solve difficult extraction, and high recognition accuracy Effect

Active Publication Date: 2020-04-21
小语智能信息科技(云南)有限公司
View PDF8 Cites 16 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The invention provides a Burmese image text recognition method based on CRNN, which is used to identify and extract Burmese text information on the image, and solves the problem that the text information in the Burmese image is difficult to extract

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
  • Myancanda image text recognition method based on CRNN
  • Myancanda image text recognition method based on CRNN
  • Myancanda image text recognition method based on CRNN

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0059] Embodiment 1: as Figure 1-2 Shown, based on the Burmese image text recognition method of CRNN, the concrete steps of described method are as follows:

[0060] Step1. Data preprocessing: Combining Burmese language features to construct training sets, test sets, and evaluation set data of long sequences and short sequences of Burmese text information images of different dynamic segments; for example, long sequence data short sequence data

[0061] Then use the Burmese Unicode sorting algorithm to mark the text information in the Burmese image. Before the training task starts, all the input Burmese image pixels are scaled to a fixed 120*32 resolution for the next deep convolutional neural network input;

[0062] Step2, feature vector sequence extraction: use the deep convolutional neural network (CNN) to extract the corresponding feature vector sequence from the input Burmese image, and use the convolutional layer and the maximum pooling layer in the deep convolutiona...

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 CRNN-based Myanmao image text recognition method, and belongs to the field of natural language processing. The method comprises the following steps: constructing a trainingset, a test set and evaluation set data of Myanmao text information images; marking text information in the Myanmao image by using a Myanmao Unicode sorting algorithm; adopting a deep convolutional neural network to extract a corresponding feature vector sequence from the input Myanhuang language image; identifying the feature vector sequence obtained in the last step by utilizing BiLSTM in a recurrent neural network RNN, and obtaining context information of the sequence, thereby obtaining probability distribution of each column of features; and calculating all label sequence probabilities byutilizing CTC, and selecting a label sequence corresponding to the maximum label sequence probability as a final prediction result of the Myanmao language of each frame in the image based on the dictionary and the mode of searching the candidate target. According to the method, the recognition of the Myanmai image text is realized, the recognition accuracy is high, and the effect is good.

Description

technical field [0001] The invention relates to a Burmese image text recognition method based on CRNN, and belongs to the technical field of natural language processing. Background technique [0002] Burmese image text recognition is a basic task in Burmese natural language research. Burmese text information on traditional images cannot be directly recognized and extracted by computer, and the text on images cannot be used for natural language processing research. The usual processing method They are all manually typed out by looking at the pictures, which is time-consuming and labor-intensive. At present, the method of combining deep learning in Chinese and English image text recognition tasks has achieved very good results, but there has been no breakthrough in the field of Burmese image text recognition. Because of the special syllable structure of Burmese, one syllable It may be composed of multiple characters and cannot be separated. Unlike English or Chinese, only a 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
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04G06N3/08
CPCG06N3/049G06N3/08G06V30/10G06N3/045G06F18/214G06F18/2415
Inventor 毛存礼谢旭阳余正涛高盛祥
Owner 小语智能信息科技(云南)有限公司
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