OCR depth learning method based on block chain mechanism, storage medium

A technology of deep learning and blockchain, applied in the field of OCR deep learning methods based on the storage medium and blockchain mechanism, can solve a lot of learning time and other problems, achieve the effect of improving learning efficiency, increasing enterprise value, and reducing learning time

Active Publication Date: 2019-02-15
福建天晴在线互动科技有限公司
View PDF2 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

And every time there is an additional font, there are tens of thousands of characters in Chinese characters alone, which requires a lot of learning 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
  • OCR depth learning method based on block chain mechanism, storage medium
  • OCR depth learning method based on block chain mechanism, storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0051] Please refer to figure 2 , this embodiment is for figure 2 A further definition of the OCR deep learning method based on the block chain mechanism is provided, including the following steps:

[0052] S1: Build a blockchain network platform.

[0053] Build a blockchain network platform, that is, a blockchain network. Each node of the blockchain network platform stores the same data and is open for use by default.

[0054] S2: Share and store OCR identification data to each node of the blockchain network.

[0055] Based on the data sharing feature of the blockchain, the OCR identification data can be obtained by any machine to achieve reusability. For example, if a new personalized font is released, if you want to be able to recognize the font, you need to perform deep learning on its OCR recognition data.

[0056] S2: Create a model consisting of each state and its corresponding current reward according to the learning data, where each state corresponds to each sub-l...

Embodiment 2

[0076] This embodiment is a specific application scenario corresponding to Embodiment 1:

[0077] The OCR deep learning method based on the blockchain mechanism can be aimed at the ever-changing personalized Chinese characters, English and Martian characters, traditional Chinese, foreign languages ​​and other unconventional characters.

[0078] For example, when a new personalized text comes out, its font is unconventional and may not be found in the known font library, so if the machine wants to learn such text recognition, deep learning is required, and using the first embodiment The method can achieve the purpose of safety and speed.

[0079] Specifically, when identifying new personalized fonts, first select characters with fewer strokes. First, according to the deep learning method, a template is given corresponding to a word (that is, the established model, deep learning is to learn by the machine after giving the established goal, and the template here refers to the es...

Embodiment 3

[0085] This embodiment corresponds to Embodiment 1 or Embodiment 2, and provides a computer-readable storage medium on which a computer program is stored. When the program is processed by a processor, it can realize the above-mentioned embodiment 1 or Embodiment 2. The steps included in the OCR deep learning method based on the blockchain mechanism. The specific steps will not be repeated here, please refer to the description of Embodiment 1 or Embodiment 2 for details.

[0086] Those of ordinary skill in the art can understand that all or part of the processes in the technical solution of the above-mentioned embodiment 1 or embodiment 2 can be realized by instructing related hardware through a computer program, and the program can be stored in a computer In the readable storage medium, when the program is executed, it may include the processes of the above-mentioned methods.

[0087] Wherein, the storage medium may be a magnetic disk, an optical disk, a read-only memory (Rea...

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 an OCR depth learning method based on a block chain mechanism. The storage medium comprises: a model composed of each state and its corresponding immediate return is created according to the learning data. The states correspond to each sub-learning data in the learning data. Private storage model to the nodes of the block chain network; The states in the model are randomly selected for training, and the new returns corresponding to the states are obtained. Obtaining a training model corresponding to the model according to the new reward optimization model; Convolution depth learning is carried out according to the training model set corresponding to a sufficient number of models. Share the results of convolutional depth learning to each node. The invention can realize resource sharing of learning data. It can effectively protect the self-learning results, improve the learning value, and help to accelerate the progress of research and development; But also guarantees user data privacy.

Description

technical field [0001] The invention relates to an OCR deep learning method, in particular to an OCR deep learning method and a storage medium based on a block chain mechanism. Background technique [0002] The current common OCR scheme is a top-down method based on sliding window full-image scanning and a bottom-up method based on underlying rules to first segment small areas and then combine them into text areas. The deep learning of OCR is based on the 5-layer CNN network of HOG feature input as the OCR recognition model. In the recognition network training, CNN needs to recognize complex characters. The problem it faces is how to obtain enough training samples that meet the diversity. Only when the training samples meet the diversity can the OCR recognition network that meets the business needs be trained. [0003] Traditional text recognition is based on template matching, which has high requirements for feature description, and it is difficult to meet the recognition ...

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): G06K9/32G06K9/62
CPCG06V20/63G06V30/10G06F18/214
Inventor 刘德建于恩涛陈伟刘萍林剑锋范剑敏林琛
Owner 福建天晴在线互动科技有限公司
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