Yazar "Cinar, Necip" seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • [ X ]
    Öğe
    An automated pothole detection via transfer learning
    (Ieee, 2022) Cinar, Necip; Kaya, Mehmet
    Potholes on the roads can cause many problems in traffic. They can cause malfunctions of vehicles, deterioration of suspension systems, additional repairs, and traffic accidents. It is very important to detect potholes quickly and with low costs for the maintenance and rehabilitation of roads. This shows that there is a need for automatic systems that can detect structural problems that may occur on the roads quickly and accurately. In this study, DenseNet121 architecture, which is a deep learning-based method, is proposed for detecting potholes in roads. With the proposed approach, it is aimed to determine whether there are potholes in the road images in the dataset. In this study, potholes on the road were detected with 99.3% accuracy using the DenseNet121 network. This success is quite high when compared to similar studies in the literature. At the same time, this dataset was run and compared with ResNet50, InceptionV3, VGG19 and InceptionResnetV2 models with the same parameters. Among these models, the highest accuracy was obtained with DenseNet121.
  • [ X ]
    Öğe
    Comparison of deep learning models for brain tumor classification using MRI images
    (Ieee, 2022) Cinar, Necip; Kaya, Buket; Kaya, Mehmet
    Brain tumor is a type of cancer that can occur in humans, sometimes with fatal consequences, or seriously affect quality of life. Detection of brain tumors using deep learning methods is a very positive development for experts. Deep learning methods enable experts to perform tumor detection and treatment easier and faster. There are many methods are used to detect tumors from brain MRI images. Among these methods, deep learning methods have made a significant improvement over other methods. In this study, it is aimed to compare the models used for tumor detection from brain MRI images within the scope of deep learning methods. For this reason, the five most commonly used convolutional neural networks for brain tumor classification are discussed. VGG19, DenseNet169, AlexNet, InceptionV3 and ResNet101 models, which are Convolutional Neural Network (CNN) architectures, were used. MR images, which underwent the same dataset and preprocessing processes, were trained with these models with the same hyper-parameters. As a result of the study, the ResNet101 model obtained the highest accuracy rate with an accuracy value of 98,6%. In addition, the VGG19 model showed a very high accuracy rate of 97.2%. Other models have accuracy values of InceptionV3 94.3%, DenseNet169 92.8%, and AlexNet 89.5%, respectively. This low success rate reveals that the architectures used in these models are not suitable for studies on MR images compared to other architectures. As a result, it has been concluded that the use of ResNet architecture for tumor detection from brain MR images is more advantageous than other models.