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

Resource allocation method based on deep learning model and terminal

A technology of deep learning and resource allocation, applied in the field of data processing, can solve problems such as slowing down, inability to effectively support the rapid expansion of bill recognition business, and slow recognition speed

Active Publication Date: 2019-05-21
上海深杳智能科技有限公司 +1
View PDF5 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The recognition accuracy of the pure manual OCR method is high enough, but there are many defects, such as the labor cost is very high, after manually recognizing the text, it is necessary to manually type the keyboard to input the text into the computer system, which will reduce the speed and introduce additional errors, and the recognition performance will be reduced. Affected by human factors such as fatigue, the recognition speed is slow, and it cannot effectively support the rapid expansion of the bill recognition business, etc.

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
  • Resource allocation method based on deep learning model and terminal
  • Resource allocation method based on deep learning model and terminal
  • Resource allocation method based on deep learning model and terminal

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0134] Such as image 3 As shown, this embodiment provides a resource allocation method based on a deep learning model, including:

[0135] S1. A plurality of different OCR deep learning models are preset; the OCR deep learning models are used to identify character segments in an image.

[0136] Wherein, the selected multiple different OCR deep learning models have certain complementarity in recognition performance. For example, when some selected deep learning models are trained, more digital training samples are used to enhance the model's training on numbers, so that the model has a relatively higher accuracy in recognizing numbers; while other selected models , can be different in the network structure and / or training of the deep learning model, so that the recognition of Chinese characters has a higher accuracy. The above two types of deep learning models are complementary in recognizing the types of characters. The complementarity between different models can be evalu...

Embodiment 2

[0231] Such as Figure 4 As shown, this embodiment provides a resource allocation terminal based on a deep learning model, including one or more processors 1 and a memory 2, the memory 2 stores a program, and is configured to be processed by the one or more Device 1 performs the following steps:

[0232] S1. A plurality of different OCR deep learning models are preset; the OCR deep learning models are used to identify character segments in an image.

[0233] Wherein, there is a certain complementarity between the multiple different OCR deep learning models. Different module parameters (such as the number of stages of the convolutional layer and the setting of the pooling layer) are set in each OCR deep learning model, or different training sample sets are used for training. Use the test sample set to test different trained OCR deep learning models, and count the accuracy of a single OCR deep learning model for recognizing character fragment images.

[0234] For example, the...

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 resource allocation method based on a deep learning model and a terminal, and belongs to the field of data processing. The method comprises: the accuracy of using differentOCR deep learning model combinations to identify the to-be-identified character segment image is estimated through a performance function capable of reflecting the relation between the confidence coefficient and the accuracy of using a plurality of OCR deep learning models to identify the character segment image at the same time; further combining different OCR deep learning model combinations with different numbers of artificial identifications; and calculating the accuracy of the combination of various machines and manual recognition, selecting a combination with the lowest cost from the combinations of all machines and manual recognition which can reach an expected accuracy target value, and distributing the number of models and the number of workers according to the selected combination. The identification cost is effectively reduced while the high identification accuracy target value is met.

Description

technical field [0001] The invention relates to a resource allocation method and a terminal based on a deep learning model, belonging to the field of data processing. Background technique [0002] In a general OCR application system, the commonly used OCR process includes obtaining an image with text, using a machine to locate the field of interest in the image, and the machine recognizes the located character field segment image and other steps. According to the accuracy requirements of the OCR application system for character recognition, it is optional to manually review the OCR recognition result to correct the wrong recognition. See the OCR system process of using the machine figure 1 . [0003] However, commercial applications such as bill recognition have high requirements on accuracy (for example, 99.9%), and it is difficult to meet the accuracy requirements by using a common OCR process. Therefore, most of the current commercial ticket recognition applications us...

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): G06K9/34G06K9/62
CPCY02D10/00
Inventor 周异何建华陈凯杜保发周曲黄征
Owner 上海深杳智能科技有限公司
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