Nameplate recognition method, computer equipment and storage medium

A nameplate, to-be-recognized technology, applied in the field of natural text recognition, can solve problems such as slowing down model recognition speed, unfavorable recognition of text content, and non-intensive text information, so as to improve text recognition rate, improve training speed, and expand the effect of receptive field

Pending Publication Date: 2021-03-16
SHANDONG UNIV
View PDF6 Cites 13 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, there are the following defects or deficiencies in this patent: (1) if the nameplate image to be detected is not horizontal, the final image of the nameplate text area is also not horizontal, which is not conducive to the next step of recognizing text content; (2) detection The PSENet used in the text area is a detection model based on segmentation. Most of the text in the nameplate image is not highly distorted, and the text information is not dense compared with the natural text image scene. Using PSENet will reduce the recognition speed of the model, while the recognition accuracy The improvement is not great; (3) the patent only covers the detection of text areas, and does not include subsequent text content recognition

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
  • Nameplate recognition method, computer equipment and storage medium
  • Nameplate recognition method, computer equipment and storage medium
  • Nameplate recognition method, computer equipment and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0055] A nameplate automatic identification method, comprising:

[0056] The image to be recognized is sent to the classification network model to obtain the direction angle of the image to be recognized, and the direction of the image to be recognized is corrected; the generation method of the classification network model includes: adjusting several nameplate pictures obtained to the level (obtaining the nameplate image used for training, by The user shoots by himself. Due to the influence of environmental factors during the image collection process, the shooting angle may not be fixed, and the obtained nameplate image will appear blurred and deformed; the nameplate part in the image to be recognized is manually adjusted to the level to obtain the training data set ); randomly rotate several fixed angles; the classification network model is obtained by training the nameplate pictures marked with fixed angles of rotation; several fixed angles include 0°, 45°, 90°, 135°, 180°, 2...

Embodiment 2

[0061] According to a kind of nameplate automatic identification method described in embodiment 1, its difference is:

[0062] Such as figure 2 As shown, the text area detection module refers to the CTPN (ConnectionistTextProposalNetwork) network. CTPN combines CNN and LSTM deep network to effectively detect horizontally distributed text in complex scenes. The text area detection module is obtained by training the nameplate image containing the label. ,Refers to:

[0063] First, use the VGG16 classification model to extract features and obtain a feature map with a size of N×C×H×W; N, C, H, and W refer to BatchSize, number of feature map channels, feature map height, and feature map width respectively; in N A 3×3 sliding window is made on the feature map of ×C×H×W, and the output of the feature map of N×(9*C)×H×W is obtained, and each point (each along the direction of height and width positions) are combined with 3×3 regional features, and the feature map of N×(9*C)×H×W is r...

Embodiment 3

[0068] According to a kind of nameplate automatic identification method described in embodiment 1 or 2, its difference is:

[0069] Such as image 3 As shown, the text recognition module includes sequentially connected STN space transformation network, feature extraction module and time convolution network;

[0070] The STN space transformation network is used to offset the influence of the image due to the incorrect shooting angle, the feature extraction module is used to extract the visual features of the text image, and the temporal convolution module extracts the text semantic features corresponding to the text image.

[0071] The text recognition module is obtained through training, including the following steps:

[0072] First of all, for the nameplate picture marked with the text image area, it is scaled to 32×320, and the STN (SpatialTransformerNetwork) spatial transformation network is used to perform adaptive affine transformation on the text image scaled to the sta...

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 relates to a nameplate recognition method, computer equipment and a storage medium, and the method comprises the steps of transmitting a to-be-recognized image to a classification network model, obtaining the direction and angle of the to-be-recognized image, and carrying out the direction correction of the to-be-recognized image; performing text region detection on the nameplate picture after direction correction through a text region detection module to obtain a nameplate picture marked with a text image region; and performing text recognition on the nameplate picture marked with the text image area through a text recognition module. The invention can automatically identify the image direction. According to the invention, the text recognition rate at different shooting angles can be improved. According to the invention, the text recognition part completely uses the convolutional network, the speed is about 1.5 times that of a CRNN text recognition network, and the accuracy is higher.

Description

technical field [0001] The invention relates to a nameplate recognition method, computer equipment and a storage medium, and belongs to the technical field of natural text recognition. Background technique [0002] At present, each engine has its own unique nameplate. During inspection and maintenance, it is necessary to determine the model through the number on the nameplate and engine parameters to facilitate management and targeted maintenance. At present, this work is mainly based on manual identification and entry system. The working environment of the engine is complex, and most of the pictures taken are affected by the environment. It is impossible to take clear and easy-to-recognize photos, and the speed and accuracy of manual identification are greatly affected. [0003] With the development of the field of artificial intelligence, the automatic recognition of text by machines has a great advantage over manual labor in terms of speed and accuracy. However, the exis...

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 Applications(China)
IPC IPC(8): G06K9/32G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V20/63G06V10/243G06V30/10G06N3/044G06N3/045G06F18/23G06F18/24G06F18/214
Inventor 段恩悦周洪超杜晓炜
Owner SHANDONG UNIV
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