Check patentability & draft patents in minutes with Patsnap Eureka AI!

An acceleration device for image text recognition

A text recognition and acceleration device technology, applied in the direction of processor architecture/configuration, can solve the problems of large amount of calculation, difficult to achieve real-time processing, long processing time, etc., and achieve the effect of high recognition accuracy

Active Publication Date: 2022-07-26
成都视海芯图微电子有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The technical problem to be solved by the present invention is: when complex network structures process text images with high precision, the amount of calculation is large, resulting in long processing time, and it is difficult to achieve the purpose of real-time processing

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
  • An acceleration device for image text recognition
  • An acceleration device for image text recognition

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0028] figure 1 It is a schematic structural diagram of an acceleration device for image and text recognition according to an embodiment. As shown in the figure, the acceleration device includes a processor 1 , a memory 2 and an image text accelerator 3 . Among them, the processor 1, the memory 2 and the image and text accelerator 3 are bidirectionally connected to each other.

[0029]The processor 1 is used to generate the execution command for the acceleration algorithm, and transmit the generated execution command to the image-text accelerator 3 to control the image-text accelerator 3 to work; at the same time, the processor 1 accesses the memory 2, acquires the image from the memory 2, and The process of executing the image text recognition algorithm includes image orientation correction, and image cropping operation is performed according to the orientation-corrected image to obtain an image containing text.

[0030] The memory 2 is used for storing pre-input image data...

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 an image-text recognition-oriented acceleration device, comprising: a processor, a memory and an image-text accelerator. Among them, the image text accelerator includes: operation array, softmax module, normalization module, embedding module and activation module. The operation array is connected with the softmax module, normalization module, embedding module and activation module, respectively. The functions of each functional module are relatively independent and work together to complete the rapid recognition of image data. Therefore, compared with the method of using complex network results for image processing, the use of this acceleration device for image recognition can ensure accuracy. It has a faster operation speed and effectively improves the efficiency of image recognition as a whole.

Description

technical field [0001] The invention relates to the technical field of image text processing, in particular to an acceleration device for image text recognition. Background technique [0002] Intelligent perception algorithms such as deep learning enable electronic devices to have accurate perception capabilities, such as gesture recognition, speech recognition, text recognition, etc., providing more accurate perception information for device management and control. A series of new methods have also appeared for text recognition in images, and the semantic perception technology for text has also developed to a certain stage. [0003] At present, the recognition of text in an image is mainly performed by detecting and cropping the text blocks in the image, and then using a text recognition method to recognize the text blocks in the cropped part of the image. In most cases, when recognizing text blocks in images, real-time text recognition can be achieved through a simple net...

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 Patents(China)
IPC IPC(8): G06T1/20
CPCG06T1/20
Inventor 张旻晋许达文
Owner 成都视海芯图微电子有限公司
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More