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

Lightweight license plate recognition method and system based on full convolutional network

A fully convolutional network, license plate recognition technology, applied in the field of lightweight license plate recognition methods and systems, can solve the problems of not supporting license plate color category recognition, unable to handle double-layer license plates of variable-length license plates, and poor robustness of modeling methods.

Pending Publication Date: 2020-10-13
SHANGHAI JIAO TONG UNIV
View PDF7 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, there are problems such as the inability to deal with variable-length license plates, double-layer license plates, and poor robustness of the modeling method.
And it does not support the recognition of license plate color categories

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
  • Lightweight license plate recognition method and system based on full convolutional network
  • Lightweight license plate recognition method and system based on full convolutional network
  • Lightweight license plate recognition method and system based on full convolutional network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0101] According to a kind of lightweight license plate recognition method based on full convolutional network provided by the present invention, comprising:

[0102] Step M1: collect and label license plate sample pictures, and divide them into training set and test set according to the ratio of 9:1;

[0103] Step M2: Build a lightweight license plate recognition network model based on full convolution based on license plate recognition;

[0104] Step M3: Determine the multi-task learning framework, and set the loss function for optimizing the parameters of the lightweight license plate recognition network model based on full convolution based on the multi-task learning framework;

[0105] Step M4: use the license plate sample pictures of the training set to train the lightweight license plate recognition network model based on full convolution until the error of the loss function is less than the preset value;

[0106] Step M5: Select the model parameters of the lightweight...

Embodiment 2

[0209] Embodiment 2 is a modification of embodiment 1

[0210] Such as figure 1 As shown, this embodiment provides a lightweight license plate recognition method based on a fully convolutional network. Contains aspects such as network model, training steps, deployment steps, etc.

[0211] Such as figure 2 As shown, this embodiment designs a license plate recognition network framework that supports accurate recognition of the character content and color category of the license plate at the same time, and widely supports single-layer, double-layer, new energy, police, military and other special types of license plates in China. Model training uses a multi-task training framework, using CELoss and CTCLoss (Graves A, Fernández S, Gomez F, et al. Connectionist temporal classification: labeling unsegmented sequence data with recurrent neural networks[C]. 2006.) as the loss function for optimizing model parameters.

[0212] data set

[0213] The datasets consist of real datasets...

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 provides a lightweight license plate recognition method and system based on a full convolutional network, and the method comprises the steps: collecting and marking license plate samplepictures, and dividing the pictures into a training set and a test set according to a preset proportion; building a lightweight license plate recognition network model based on full convolution; determining a multi-task learning framework and a set loss function; training a full convolution-based lightweight license plate recognition network model by using the license plate sample pictures of thetraining set until the error of the loss function is smaller than a preset value; selecting full-convolution-based lightweight license plate recognition network model parameters stored in different stages in the training process, testing the comprehensive performance of a full-convolution-based lightweight license plate recognition network under different parameters by using license plate sample pictures in a test set, and fixing the parameter with the highest accuracy as a final parameter of the model. According to the method, the sequence information is modeled by adopting the full convolutional network, so that the model is easy to realize in parallel, fewer computing resources are required in the reasoning stage, and the time delay is lower.

Description

technical field [0001] The present invention relates to digital image processing technology and optical character recognition (OCR) technology in the field of computer vision, in particular, to a light-weight license plate recognition method and system based on a fully convolutional network. Background technique [0002] With the rapid development of economy and society, the number of motor vehicles is also increasing. Realizing automatic identification of vehicle identities can improve vehicle management efficiency and reduce labor costs. Therefore, license plate recognition technology has become a research hotspot in recent years. The current common license plate recognition technology can be divided into recognition technology based on character segmentation and recognition technology based on end-to-end network. [0003] The recognition technology based on character segmentation is to divide the continuous license plate content into multiple single characters, and then...

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): G06K9/46G06K9/62
CPCG06V10/56G06V20/625G06V30/10G06F18/2415G06F18/214
Inventor 孙锬锋蒋兴浩李季许可
Owner SHANGHAI JIAO TONG UNIV
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