Thesis Open Access

# ETHIOPIAN CAR LICENSE PLATE RECOGNITION USING DEEP LEARNING

ERDEY SYOUM

### DataCite XML Export

<?xml version='1.0' encoding='utf-8'?>
<creators>
<creator>
<creatorName>ERDEY SYOUM</creatorName>
<affiliation>ADDIS ABABA SCIENCE AND TECHNOLOGY UNIVERSITY</affiliation>
</creator>
</creators>
<titles>
<title>ETHIOPIAN CAR LICENSE PLATE RECOGNITION USING DEEP LEARNING</title>
</titles>
<publisher>National Academic Digital Repository of Ethiopia</publisher>
<publicationYear>2020</publicationYear>
<subjects>
<subject>Automatic Vehicle Identification Convolutional Neural Network Deep Learning Object Detection Optical Character Recognition</subject>
</subjects>
<contributors>
<contributor contributorType="Supervisor">
<contributorName>Sreenivasa Rao.vuda (PhD)</contributorName>
<affiliation>ADDIS ABABA SCIENCE AND TECHNOLOGY UNIVERSITY</affiliation>
</contributor>
</contributors>
<dates>
<date dateType="Issued">2020-01-16</date>
</dates>
<language>en</language>
<resourceType resourceTypeGeneral="Text">Thesis</resourceType>
<alternateIdentifiers>
</alternateIdentifiers>
<relatedIdentifiers>
</relatedIdentifiers>
<rightsList>
<rights rightsURI="info:eu-repo/semantics/openAccess">Open Access</rights>
</rightsList>
<descriptions>
<description descriptionType="Abstract">&lt;p&gt;The focus of this research is to develop a system that assist humans in reading car&lt;br&gt;
license plate. Such a study is important as the number of traffic on roads becomes&lt;br&gt;
increasing constantly, the manual process in car license plate recognition becomes a&lt;br&gt;
serious problem for traffic management system which not only detect and track a&lt;br&gt;
vehicle but also identify it. Initially a dataset that contains 930 car images was prepared&lt;br&gt;
for model comparison purpose. Two object detection algorithms (Faster R-CNN and&lt;br&gt;
SSD) were trained and tested on the same dataset using the same model to select the&lt;br&gt;
best candidate. The metrics for the comparison were accuracy, average prediction time,&lt;br&gt;
and total training time taken. It was found that Faster R-CNN gives high accuracy, short&lt;br&gt;
average prediction time, and short total training time. After that additional car and&lt;br&gt;
cropped license plate images were added to the prepared dataset and based on this, two&lt;br&gt;
object detection networks were trained using Faster R-CNN one for plate detection and&lt;br&gt;
another for character recognition on the detected plate. The proposed approach has been&lt;br&gt;
tested on test set and later collected images of national license plate of Ethiopia. Both&lt;br&gt;
the trained models were achieved a high accuracy which is 99 and 98.89 mAP over 0.5&lt;br&gt;
IoU for plate detection and character recognition respectively and takes on average 12s&lt;br&gt;
to complete the recognition of a license plate. The study could be further investigated&lt;br&gt;
on other countries.&lt;/p&gt;</description>
</descriptions>
</resource>

202
268
views