{"id":{"repo_id":"ankara","oai_identifier":"oai:dspace.ankara.edu.tr:20.500.12575/82728"},"canonical_url":"https://search.dev.ndltd.org/etd/ankara/oai:dspace.ankara.edu.tr:20.500.12575/82728","repository":{"repo_id":"ankara","name":"University of Ankara","base_url":"https://dspace.ankara.edu.tr/server/oai/request"},"display":{"title":"İmge içeriği tabanlı nesne sınıflandırma","abstract":"Image processing technologies are used in many fields which are defense, medical, communication and automotive. The use of artificial intelligence in image processing technologies has led to speed up autonomous vehicle studies in the automotive sector. In this study, it is aimed at an artificial neural network that was originally designed to automatically detect and recognize traffic signs on the images coming to the network. For this purpose, supervised deep learning technique was used to identify, classify and mark the signs belongs to 10 different classes. YZNet network, which is designed within the scope of the thesis, consists of 16 layers including input, output and hidden layers. Firstly, YZNet was trained using the German Traffic Sign Recognition Benchmark, using the Convolution Neural Network (CNN) method with supervised learning for classification. The number of pictures used for education is 26640. In this data set, the accuracy of the YZNet network was 93.05%. YZNet network designed in the study and AlexNet and Cifar10Net networks are available in the literature were retrained using the Regional Convolution Neural Network (R-CNN) method to detect and recognize traffic signs in the image and mark them with a bounding box and put the label on them. The German Traffic Sign Detection Benchmark was used for training and testing with the R-CNN method. 600 pictures in the data set were used for training and 300 pictures for testing. In the study the lowest false alarm rates were obtained as 0.2 in AlexNet, 0.25 in Cifar10Net and 0.72 in YZNet. The highest accuracy values were obtained as 98.33% for AlexNet, 98.33% for Cifar10Net and 96.95% for YZNet.","abstract_html":"Image processing technologies are used in many fields which are defense, medical, communication and automotive. The use of artificial intelligence in image processing technologies has led to speed up autonomous vehicle studies in the automotive sector. In this study, it is aimed at an artificial neural network that was originally designed to automatically detect and recognize traffic signs on the images coming to the network. For this purpose, supervised deep learning technique was used to identify, classify and mark the signs belongs to 10 different classes. YZNet network, which is designed within the scope of the thesis, consists of 16 layers including input, output and hidden layers. Firstly, YZNet was trained using the German Traffic Sign Recognition Benchmark, using the Convolution Neural Network (CNN) method with supervised learning for classification. The number of pictures used for education is 26640. In this data set, the accuracy of the YZNet network was 93.05%. YZNet network designed in the study and AlexNet and Cifar10Net networks are available in the literature were retrained using the Regional Convolution Neural Network (R-CNN) method to detect and recognize traffic signs in the image and mark them with a bounding box and put the label on them. The German Traffic Sign Detection Benchmark was used for training and testing with the R-CNN method. 600 pictures in the data set were used for training and 300 pictures for testing. In the study the lowest false alarm rates were obtained as 0.2 in AlexNet, 0.25 in Cifar10Net and 0.72 in YZNet. The highest accuracy values were obtained as 98.33% for AlexNet, 98.33% for Cifar10Net and 96.95% for YZNet.","abstract_has_math":false,"creators":["Kalkan, Yeşim"],"institution":"Fen Bilimleri Enstitüsü","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":"Elektronik Mühendisliği","school":null,"contributors":[],"advisors":["Telatar, Ziya"],"committee_chairs":[],"committee_members":[],"year":2020,"date_issued":"2020","date_published":"2020","updated_at":"2026-08-21T16:41:48Z","subjects":["görüntü işleme","derin öğrenme","Tanıma"],"languages":["tr"],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/20.500.12575/82728","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"source_record":{"url":"https://dspace.ankara.edu.tr/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Adspace.ankara.edu.tr%3A20.500.12575%2F82728","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Telatar, Ziya"]},{"key":"dc:contributor.department","label":"Department","values":["Elektronik Mühendisliği"]},{"key":"dc:creator","label":"Author","values":["Kalkan, Yeşim"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2022-07-07T12:02:34Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2022-07-07T12:02:34Z"]},{"key":"dc:date.issued","label":"Date","values":["2020"]},{"key":"dc:publisher","label":"Institution","values":["Fen Bilimleri Enstitüsü"]},{"key":"dc:type","label":"Dc Type","values":["masterThesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["görüntü işleme","derin öğrenme","Tanıma"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["tr"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["http://hdl.handle.net/20.500.12575/82728"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Image processing technologies are used in many fields which are defense, medical, communication and automotive. The use of artificial intelligence in image processing technologies has led to speed up autonomous vehicle studies in the automotive sector. In this study, it is aimed at an artificial neural network that was originally designed to automatically detect and recognize traffic signs on the images coming to the network. For this purpose, supervised deep learning technique was used to identify, classify and mark the signs belongs to 10 different classes. YZNet network, which is designed within the scope of the thesis, consists of 16 layers including input, output and hidden layers. Firstly, YZNet was trained using the German Traffic Sign Recognition Benchmark, using the Convolution Neural Network (CNN) method with supervised learning for classification. The number of pictures used for education is 26640. In this data set, the accuracy of the YZNet network was 93.05%. YZNet network designed in the study and AlexNet and Cifar10Net networks are available in the literature were retrained using the Regional Convolution Neural Network (R-CNN) method to detect and recognize traffic signs in the image and mark them with a bounding box and put the label on them. The German Traffic Sign Detection Benchmark was used for training and testing with the R-CNN method. 600 pictures in the data set were used for training and 300 pictures for testing. In the study the lowest false alarm rates were obtained as 0.2 in AlexNet, 0.25 in Cifar10Net and 0.72 in YZNet. The highest accuracy values were obtained as 98.33% for AlexNet, 98.33% for Cifar10Net and 96.95% for YZNet."]},{"key":"dc:title","label":"Title","values":["İmge içeriği tabanlı nesne sınıflandırma"]}]}],"canonical_facts":{"dc:contributor.advisor":["Telatar, Ziya"],"dc:contributor.department":["Elektronik Mühendisliği"],"dc:creator":["Kalkan, Yeşim"],"dc:date.accessioned":["2022-07-07T12:02:34Z"],"dc:date.available":["2022-07-07T12:02:34Z"],"dc:date.issued":["2020"],"dc:description.abstract":["Image processing technologies are used in many fields which are defense, medical, communication and automotive. The use of artificial intelligence in image processing technologies has led to speed up autonomous vehicle studies in the automotive sector. In this study, it is aimed at an artificial neural network that was originally designed to automatically detect and recognize traffic signs on the images coming to the network. For this purpose, supervised deep learning technique was used to identify, classify and mark the signs belongs to 10 different classes. YZNet network, which is designed within the scope of the thesis, consists of 16 layers including input, output and hidden layers. Firstly, YZNet was trained using the German Traffic Sign Recognition Benchmark, using the Convolution Neural Network (CNN) method with supervised learning for classification. The number of pictures used for education is 26640. In this data set, the accuracy of the YZNet network was 93.05%. YZNet network designed in the study and AlexNet and Cifar10Net networks are available in the literature were retrained using the Regional Convolution Neural Network (R-CNN) method to detect and recognize traffic signs in the image and mark them with a bounding box and put the label on them. The German Traffic Sign Detection Benchmark was used for training and testing with the R-CNN method. 600 pictures in the data set were used for training and 300 pictures for testing. In the study the lowest false alarm rates were obtained as 0.2 in AlexNet, 0.25 in Cifar10Net and 0.72 in YZNet. The highest accuracy values were obtained as 98.33% for AlexNet, 98.33% for Cifar10Net and 96.95% for YZNet."],"dc:identifier.uri":["http://hdl.handle.net/20.500.12575/82728"],"dc:language.iso":["tr"],"dc:publisher":["Fen Bilimleri Enstitüsü"],"dc:subject":["görüntü işleme","derin öğrenme","Tanıma"],"dc:title":["İmge içeriği tabanlı nesne sınıflandırma"],"dc:type":["masterThesis"]},"updated_at":"2026-08-21T16:41:48Z"}