4 results
Search Results
Now showing 1 - 4 of 4
Article Citation - WoS: 17Citation - Scopus: 31Deep Learning-Based Vehicle Classification for Low Quality Images(Mdpi, 2022) Tas, Sumeyra; Sari, Ozgen; Dalveren, Yaser; Pazar, Senol; Kara, Ali; Derawi, MohammadThis study proposes a simple convolutional neural network (CNN)-based model for vehicle classification in low resolution surveillance images collected by a standard security camera installed distant from a traffic scene. In order to evaluate its effectiveness, the proposed model is tested on a new dataset containing tiny (100 x 100 pixels) and low resolution (96 dpi) vehicle images. The proposed model is then compared with well-known VGG16-based CNN models in terms of accuracy and complexity. Results indicate that although the well-known models provide higher accuracy, the proposed method offers an acceptable accuracy (92.9%) as well as a simple and lightweight solution for vehicle classification in low quality images. Thus, it is believed that this study might provide useful perception and understanding for further research on the use of standard low-cost cameras to enhance the ability of the intelligent systems such as intelligent transportation system applications.Article Citation - WoS: 3Citation - Scopus: 4Modelling and Design of Pre-Equalizers for a Fully Operational Visible Light Communication System(Mdpi, 2023) Bostanoglu, Murat; Dalveren, Yaser; Catak, Ferhat Ozgur; Kara, AliNowadays, Visible Light Communication (VLC) has gained much attention due to the significant advancements in Light Emitting Diode (LED) technology. However, the bandwidth of LEDs is one of the important concerns that limits the transmission rates in a VLC system. In order to eliminate this limitation, various types of equalization methods are employed. Among these, using digital pre-equalizers can be a good choice because of their simple and reusable structure. Therefore, several digital pre-equalizer methods have been proposed for VLC systems in the literature. Yet, there is no study in the literature that examines the implementation of digital pre-equalizers in a realistic VLC system based on the IEEE 802.15.13 standard. Hence, the purpose of this study is to propose digital pre-equalizers for VLC systems based on the IEEE 802.15.13 standard. For this purpose, firstly, a realistic channel model is built by collecting the signal recordings from a real 802.15.13-compliant VLC system. Then, the channel model is integrated into a VLC system modeled in MATLAB. This is followed by the design of two different digital pre-equalizers. Next, simulations are conducted to evaluate their feasibility in terms of the system's BER performance under bandwidth-efficient modulation schemes, such as 64-QAM and 256-QAM. Results show that, although the second pre-equalizer provides lower BERs, its design and implementation might be costly. Nevertheless, the first design can be selected as a low-cost alternative to be used in the VLC system.Article Citation - WoS: 29Citation - Scopus: 33On the Performance of Variational Mode Decomposition-Based Radio Frequency Fingerprinting of Bluetooth Devices(Mdpi, 2020) Aghnaiya, Alghannai; Dalveren, Yaser; Kara, AliRadio frequency fingerprinting (RFF) is one of the communication network's security techniques based on the identification of the unique features of RF transient signals. However, extracting these features could be burdensome, due to the nonstationary nature of transient signals. This may then adversely affect the accuracy of the identification of devices. Recently, it has been shown that the use of variational mode decomposition (VMD) in extracting features from Bluetooth (BT) transient signals offers an efficient way to improve the classification accuracy. To do this, VMD has been used to decompose transient signals into a series of band-limited modes, and higher order statistical (HOS) features are extracted from reconstructed transient signals. In this study, the performance bounds of VMD in RFF implementation are scrutinized. Firstly, HOS features are extracted from the band-limited modes, and then from the reconstructed transient signals directly. Performance comparison due to both HOS feature sets is presented. Moreover, the lower SNR bound within which the VMD can achieve acceptable accuracy in the classification of BT devices is determined. The approach has been tested experimentally with BT devices by employing a Linear Support Vector Machine (LSVM) classifier. According to the classification results, a higher classification performance is achieved (similar to 4% higher) at lower SNR levels (-5-5 dB) when HOS features are extracted from band-limited modes in the implementation of VMD in RFF of BT devices.Article Citation - WoS: 3Citation - Scopus: 6Investigating the Impact of Two Major Programming Environments on the Accuracy of Deep Learning-Based Glioma Detection From Mri Images(Mdpi, 2023) Yilmaz, Vadi Su; Akdag, Metehan; Dalveren, Yaser; Doruk, Resat Ozgur; Kara, Ali; Soylu, AhmetBrain tumors have been the subject of research for many years. Brain tumors are typically classified into two main groups: benign and malignant tumors. The most common tumor type among malignant brain tumors is known as glioma. In the diagnosis of glioma, different imaging technologies could be used. Among these techniques, MRI is the most preferred imaging technology due to its high-resolution image data. However, the detection of gliomas from a huge set of MRI data could be challenging for the practitioners. In order to solve this concern, many Deep Learning (DL) models based on Convolutional Neural Networks (CNNs) have been proposed to be used in detecting glioma. However, understanding which CNN architecture would work efficiently under various conditions including development environment or programming aspects as well as performance analysis has not been studied so far. In this research work, therefore, the purpose is to investigate the impact of two major programming environments (namely, MATLAB and Python) on the accuracy of CNN-based glioma detection from Magnetic Resonance Imaging (MRI) images. To this end, experiments on the Brain Tumor Segmentation (BraTS) dataset (2016 and 2017) consisting of multiparametric magnetic MRI images are performed by implementing two popular CNN architectures, the three-dimensional (3D) U-Net and the V-Net in the programming environments. From the results, it is concluded that the use of Python with Google Colaboratory (Colab) might be highly useful in the implementation of CNN-based models for glioma detection. Moreover, the 3D U-Net model is found to perform better, attaining a high accuracy on the dataset. The authors believe that the results achieved from this study would provide useful information to the research community in their appropriate implementation of DL approaches for brain tumor detection.
