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

Deep neural network natural scene text detection method suitable for dense texts

A deep neural network and text detection technology, which is applied in the field of deep neural network natural scene text detection, can solve problems such as poor dense text detection results and poor lighting conditions

Pending Publication Date: 2021-12-21
SICHUAN UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

(5) Poor lighting conditions and varying degrees of occlusion
However, this method has poor detection results for dense text

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
  • Deep neural network natural scene text detection method suitable for dense texts
  • Deep neural network natural scene text detection method suitable for dense texts
  • Deep neural network natural scene text detection method suitable for dense texts

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0017] The present invention will be further described below in conjunction with accompanying drawing:

[0018] A deep neural network natural scene text detection method suitable for dense text can be divided into the following steps:

[0019] (1) In the feature extraction layer, construct a hole convolution module;

[0020] (2) According to the feature extraction layer obtained in step (1), construct a feature fusion layer that introduces a corner point attention mechanism;

[0021] (3) According to the feature extraction layer that step (1) obtains and the feature fusion layer that step (2) obtains, construct network output layer, obtain a kind of text detection deep neural network model applicable to dense text;

[0022] (4) Utilize training data set, design and introduce the loss function of class weight factor and sample difficulty weight factor to train the depth neural network model constructed in step (3);

[0023] (5) Input the natural scene image into the deep neur...

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 deep neural network natural scene text detection method suitable for dense texts. The method mainly comprises the following steps: constructing a cavity convolution module in a feature extraction layer; constructing a feature fusion layer introducing a corner attention mechanism according to the feature extraction layer obtained in the previous step; creating a network output layer according to the feature extraction layer obtained in the first step and the feature fusion layer obtained in the second step to acquire a text detection deep neural network model suitable for dense texts; designing a loss function which introduces a category weight factor and a sample difficulty weight factor by using the training data set to train the deep neural network model constructed in the previous step; and inputting a natural scene image into the deep neural network model trained in the previous step to obtain a text detection image in the image. The deep neural network natural scene text detection method suitable for dense texts is good in detection effect of the dense texts in the natural scene and is effective.

Description

technical field [0001] The invention relates to text detection technology, in particular to a deep neural network natural scene text detection method applicable to dense text, and belongs to the field of natural scene text detection. Background technique [0002] Text detection and recognition in natural scenes is considered to be one of the most challenging problems in the field of target detection. It has a large number of applications in many machine vision fields such as image processing, driverless driving, document analysis, and natural language processing. . Its detection method is mainly divided into two parts: text detection and text recognition. Compared with the target detection of general objects, there are many difficulties in text detection in complex scenes: (1) The color, font, and scale of text lines in the scene are diverse and less relevant. (2) The background is diverse. In natural scenes, the background of text lines is arbitrary and may be affected b...

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): G06F40/205G06N3/04G06N3/08
CPCG06F40/205G06N3/08G06N3/045
Inventor 卿粼波牟森陈洪刚何小海王思怡
Owner SICHUAN UNIV