Unlock instant, AI-driven research and patent intelligence for your innovation.

Scale Adaptive Natural Scene Text Detection Method Based on Convolutional Neural Network

A convolutional neural network and scale-adaptive technology, applied in the field of computer vision, achieves fast text positioning, high practical application value, and accurate text positioning

Active Publication Date: 2021-11-05
DALIAN UNIV OF TECH
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Addresses the shortcomings of current methods

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
  • Scale Adaptive Natural Scene Text Detection Method Based on Convolutional Neural Network
  • Scale Adaptive Natural Scene Text Detection Method Based on Convolutional Neural Network
  • Scale Adaptive Natural Scene Text Detection Method Based on Convolutional Neural Network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0048] The specific implementation manners of the present invention will be further described below in conjunction with the accompanying drawings and technical solutions.

[0049] The invention builds a network model based on the Caffe framework, and directly predicts the position coordinates of the text in the natural scene image end-to-end.

[0050] Step 1: Data preparation, mark the text box in the scene: (x, y, w, h), where x, y are the abscissa and ordinate of the center point of the text box, w, h are the width of the text box respectively and high.

[0051] Step 2: Build a deep network structure, the hierarchical structure is as follows figure 2 shown. Use VGG16 as the base network, delete all layers after the Conv4_3norm layer. A scale regression layer is added after the Conv4_3norm layer, and a single-channel scale map is obtained from the scale regression layer. The scale map has the same height and width as the Conv4_3norm layer. The value of each pixel in 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 belongs to the technical field of computer vision and provides a scale-adaptive natural scene text detection method based on a convolutional neural network. This method first uses the scale regression layer to learn the scale of the text in the scene image, and then dynamically adjusts the size of the prior frame and the receptive field according to the scale of the text. The network structure designed by this method is end-to-end, which can directly locate the position of the text box in the image. Robust, accurate and fast text positioning can be realized by adopting the invention, and has high practical application value.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to a scale-adaptive natural scene text detection method based on a convolutional neural network. Background technique [0002] Text detection in natural scenes has received increasing attention in the field of computer vision due to its wide application in many practical applications such as document analysis, scene understanding, robot navigation, image retrieval, etc. Unlike document images, scene images have complex backgrounds and are easily disturbed by uncontrollable environmental factors. In addition, images in natural scenes also have variability in text size, layout, and color, so text detection in natural scenes remains an open and challenging problem. [0003] In recent years, inspired by the great progress of deep learning methods for general object detection, many deep learning based methods have achieved good results on text detection tasks. Max Jaderberg e...

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): G06K9/00G06N3/04
CPCG06V30/412G06N3/045
Inventor 李豪杰袁琪张炳旺王智慧刘华
Owner DALIAN UNIV OF TECH