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

A license plate detection method based on negative sample data value resampling

A technology of license plate detection and negative samples, which is applied in the field of computer vision, can solve problems such as the imbalance of positive and negative samples, and achieve the effect of improving robustness and alleviating the imbalance of positive and negative samples

Active Publication Date: 2020-07-17
ZHEJIANG UNIV
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Therefore, when training license plate and non-license plate classifiers, there is a serious imbalance between positive and negative samples, which will make the trained classifier have a preference for predicting negative samples

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
  • A license plate detection method based on negative sample data value resampling
  • A license plate detection method based on negative sample data value resampling
  • A license plate detection method based on negative sample data value resampling

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0026] The technical solution of the present invention is clearly and completely explained and described below.

[0027] The invention proposes a license plate detection method based on negative sample data value resampling, which can obtain a sample set beneficial to training by grouping and resampling from a large number of negative samples. Including the following steps:

[0028] Step 1. Collect the positive and negative samples required for license plate detection. The number of positive samples in the training set, verification set and test set is 3000, 1000 and 1000 respectively; the corresponding number of negative samples is 30000, 1000 and 1000. All sample image patches are scaled to 224×224.

[0029] Step 2, train the initial classifier. Randomly select 1000 negative samples, and train the AlexNet structure classifier together with the positive samples. The output uses a Softmax unit with 2 nodes. The model learning rate is set to 0.001, and the batch size is 64. ...

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

A license plate detection method based on negative sample data value resampling, including: 1) collecting license plate pictures, cutting out the license plate area as a positive sample, randomly cutting out image blocks in the non-license plate area as negative samples, dividing the training set, and verifying 2) Randomly select samples with the same number of positive samples from the negative samples in the training set to train the initial classifier, predict all the negative samples in the training set, group them according to the probability values ​​of the predicted negative samples, and Ensure that the sample size of each group except the last group is the same as the number of positive samples; 3) Retrain the classifier for each group of negative sample data and positive sample data, and calculate the information gain on the verification set to measure the data of each group of negative samples 4) Calculate the weight according to the data value of each group of negative samples, re-randomly sample from each group to form a new negative sample training set, train the final classifier with the positive samples, and use the test set to evaluate its effect.

Description

technical field [0001] The invention belongs to the field of computer vision, and aims at the problem of unbalanced positive and negative samples in a license plate detection scene, and proposes a license plate detection method based on negative sample data value resampling. Background technique [0002] In the license plate detection scene, usually only 1 or 2 license plates appear in a picture containing the license plate, and only a small number of image blocks containing the license plate area can be cropped out as positive samples. Compared with the positive samples, the collection of negative samples is relatively easy, and a large number of negative samples can be generated by randomly cropping image blocks in the remaining area except the license plate. Therefore, when training license plate and non-license plate classifiers, there is a serious imbalance between positive and negative samples, which will make the trained classifier have a preference for predicting neg...

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): G06K9/32G06K9/62
CPCG06V20/63G06V20/625G06F18/24G06F18/214
Inventor 宋明黎雷杰宋杰
Owner ZHEJIANG UNIV