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

Character detection method and device, electronic equipment and storage medium

A text detection and text frame technology, applied in the field of image processing, can solve problems such as inaccurate fitting and inaccurate detection results, and achieve the effect of improving accuracy and fitting accuracy

Active Publication Date: 2019-11-12
BEIJING SANKUAI ONLINE TECH CO LTD
View PDF5 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The existing technical solution only performs single-step regression on the text box, and the position of the box, especially the boundary of the box, is often inaccurately fitted, resulting in inaccurate detection results

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 detection method and device, electronic equipment and storage medium
  • Character detection method and device, electronic equipment and storage medium
  • Character detection method and device, electronic equipment and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0067] A text detection method disclosed in this embodiment, such as figure 2 As shown, the method includes: step 210 to step 260.

[0068] Step 210: Extract multi-level features from the image to be detected through the convolutional neural network.

[0069] Among them, the convolutional neural network is the bottom-up path in the feature pyramid network. The feature map extracted by the convolution kernel calculation of the image to be detected is getting smaller and smaller. There are usually many layers that produce output images of the same size. These layers are in In the same network stage, each stage is a pyramid level, and the output of the last layer of each stage is used as the reference feature atlas to obtain multi-level features, such as image 3 Shown. Multi-level features are represented in the form of multiple reference feature maps. One pyramid level extracts one reference feature map, and multiple pyramid levels extract multiple reference feature maps. The multi...

Embodiment 2

[0117] A text detection device disclosed in this embodiment, such as Figure 4 As shown, the text detection device 400 includes:

[0118] The feature extraction module 410 is configured to extract multi-level features from the image to be detected through a convolutional neural network;

[0119] The initial regression module 420 is configured to perform regression on the multi-level features and determine the initial text box feature in the multi-level features;

[0120] The pyramid feature generation module 430 is configured to generate pyramid features according to the multi-level features;

[0121] The new feature construction module 440 is configured to construct a new text frame feature according to the initial text frame feature and the pyramid feature;

[0122] The secondary regression module 450 is configured to perform a secondary regression on the new text box feature according to the initial text box feature to obtain the text box feature;

[0123] The text box merging module...

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 embodiment of the invention discloses a character detection method and device, electronic equipment and a storage medium. The method comprises the steps of extracting multi-level features of a to-be-detected image through a convolutional neural network; carrying out regression on the multi-level features, and determining initial text box features in the multi-level features are determined; generating pyramid features according to the multi-level features; constructing a new text box feature according to the initial text box feature and the pyramid feature; performing quadratic regression on the new textbox feature according to the initial textbox feature to obtain a textbox feature; and combining the textbox features to obtain a text detection result in the to-be-detected image. According to the embodiment of the invention, on the basis of performing regression on the multi-level features to obtain the initial textbox features, secondary regression is performed on the new textbox features, so that the fitting accuracy of the textbox boundary is improved, and the accuracy of a text detection result is improved.

Description

Technical field [0001] This application relates to the field of image processing technology, and in particular to a text detection method, device, electronic equipment and storage medium. Background technique [0002] In the prior art, the feature pyramid network is applied to the Single Shot MultiBox Detector (SSD) framework for text detection. Such as figure 1 As shown, when using the feature pyramid network for text detection, the specific steps are: the original convolutional neural network uses a top-down path and lateral connections to construct a feature pyramid network; the feature pyramid network features each layer figure 1 Used for text box classification (whether text) and regression (the center position of the box and the corresponding size); all detection boxes are suppressed by non-maximum values ​​to obtain the final text detection result. [0003] The prior art solution only performs single-step regression on the text box, and the position of the box, especially th...

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): G06K9/46G06N3/04
CPCG06V10/44G06V30/10G06N3/045
Inventor 刘曦张睿
Owner BEIJING SANKUAI ONLINE TECH CO LTD
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