Construction method of natural scene character region detection model based on no anchor point

A technology of natural scenes and text areas, applied in the field of image processing, can solve the problems of low accuracy and achieve the effect of returning to stability, suppressing interference information, and enhancing positive information

Pending Publication Date: 2020-12-29
南昌慧亦臣科技有限公司
View PDF0 Cites 10 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The present invention provides a method for constructing an anchor-free natural scene text area detec

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
  • Construction method of natural scene character region detection model based on no anchor point
  • Construction method of natural scene character region detection model based on no anchor point
  • Construction method of natural scene character region detection model based on no anchor point

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0043] The present invention provides a method for constructing a character region detection model in a natural scene without an anchor point, which is applied to occasions with high real-time requirements, and ensures high accuracy while maintaining a fast detection speed.

[0044] figure 1 It is a flow chart of the construction method of the anchor-free natural scene text area detection model based on the present invention, figure 2 It is a network structure diagram of the method for constructing a natural scene text area detection model based on no anchor points in the present invention, combined with figure 1 and figure 2 As shown, the construction method of the anchor-free natural scene text region detection model of the present invention includes,

[0045] Step S100, collect a data set containing text images for natural scenes, the data set includes a training image set T train and detection image set T test .

[0046] In step S200, the natural image is used as an...

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 a construction method of a natural scene character region detection model based on no anchor point, and the method comprises the steps: introducing a convolution score for predicting the inclination angle of a boundary frame in a detection mode based on pixels, so as to detect an inclined character in a natural scene; adding deformable convolution DCN into some layers of anetwork backbone, so the ability of the network to express specific features of a text instance is improved, and the receptive field of a text target shape is more flexible; introducing an attention module into the network, so extracted features are filtered, positive information is enhanced, and interference information is suppressed. According to the method, the classification loss, the regression loss CIoU Loss, the centrality loss and the angle loss are used as a joint loss function, so the detection precision is improved, the target frame regression becomes more stable, and meanwhile, a higher convergence speed is achieved.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for constructing an anchor-free natural scene text region detection model. Background technique [0002] Text region detection is a research hotspot in the field of computer vision. It aims to detect the position of the text in the natural scene image for the next step of recognition, so as to convert the image into real text information that can be processed by the computer. The text in natural scene images usually has large differences in fonts, combination methods, and text sizes, and natural scene images also have great uncertainties in terms of light intensity, resolution, image noise, and shooting angles. , these complex factors greatly increase the difficulty of text region detection in natural scenes. [0003] A commonly used method for text area detection in natural scenes is the method based on bounding box regression. The method based on bounding box...

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): G06K9/00G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V30/40G06V10/44G06V30/10G06N3/045G06F18/241
Inventor 徐亦飞王冕王爱臣严汤文王优李斌尉萍萍肖志峰
Owner 南昌慧亦臣科技有限公司
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