Yazar "Dogan, Yahya" seçeneğine göre listele
Listeleniyor 1 - 13 / 13
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe A New Approach for Plotting Raster Based Image Files(IEEE, 2014) Dogan, Yahya; Atas, Musa; Ozdemir, CuneytIn this study a new approach that is used for plotting raster based image files, is proposed. Proposed method is tested on the prototype plotter which is designed as a Cartesian-robot architecture. Work list is produced by the proposed edge tracking algorithm on the canny edge detector based images. Outputs are generated from the work list as a converted atomic move commands which can be interpreted by Cartesian-robot. Edge images by using conventional vectorization methods are compared with outputs of the proposed algorithm. It is observed that proposed method is superior to other techniques, in terms of representing original images.Öğe A new approach for plotting raster based image files(IEEE Computer Society, 2014) Dogan, Yahya; Atas, Musa; Ozdemir, CuneytIn this study a new approach that is used for plotting raster based image files, is proposed. Proposed method is tested on the prototype plotter which is designed as a Cartesian-robot architecture. Work list is produced by the proposed edge tracking algorithm on the canny edge detector based images. Outputs are generated from the work list as a converted atomic move commands which can be interpreted by Cartesian-robot. Edge images by using conventional vectorization methods are compared with outputs of the proposed algorithm. It is observed that proposed method is superior to other techniques, in terms of representing original images. © 2014 IEEE.Öğe A New Global Pooling Method for Deep Neural Networks: Global Average of Top-K Max(Int Information & Engineering Technology Assoc, 2023) Dogan, YahyaGlobal Pooling (GP) is one of the important layers in deep neural networks. GP significantly reduces the number of model parameters by summarizing the feature maps and enables a reduction in the computational cost of training. The most commonly used GP methods are global max pooling (GMP) and global average pooling (GAP). The GMP method produces successful results in experimental studies but has a tendency to overfit training data and may not generalize well to test data. On the other hand, the GAP method takes into account all activations in the pooling region, which reduces the effect of high activation areas and causes a decrease in model performance. In this study, a GP method called global average of top-k max pooling (GAMP) is proposed, which returns the average of the highest k activations in the feature map and allows for mixing the two methods mentioned. The proposed method is compared quantitatively with other GP methods using different models, i.e., Custom and VGG16-based and different datasets, i.e., CIFAR10 and CIFAR100. The experimental results show that the proposed GAMP method provides better image classification accuracy than the other GP methods. When the Custom model is used, the proposed GAMP method provides a classification accuracy of 1.29% higher on the CIFAR10 dataset and 1.72% higher on the CIFAR100 dataset compared to the method with the closest performance.Öğe A new local pooling approach for convolutional neural network: local binary pattern(Springer, 2023) Ozdemir, Cuneyt; Dogan, Yahya; Kaya, YilmazThe pooling layer used in CNN models aims to reduce the resolution of image/feature maps while retaining their distinctive information, reducing computation time and enabling deeper models. Max and average pooling methods are frequently used in CNN models due to their computational efficiency; however, these methods discard the position information of the pixels. In this study, we proposed an LBP-based pooling method that generates a neighborhood-based output for any pixel, reflecting the correlation between pixels in the local area. Our proposed method reduces information loss since it considers the neighborhood and size of the pixels in the pooling region. Experimental studies were performed on four public datasets to assess the effectiveness of the LBP pooling method. In experimental studies, a toy CNN model and various transfer learning models were utilized in conducting test operations. The proposed method provided improvements of 1.56% for Fashion MNIST, 0.22% for MNIST, 3.95% for CIFAR10, and 5% for CIFAR100 dataset using the toy model. In the experimental studies conducted using the transfer learning model, performance improvements of 6.99(-/+)(0.74) and 8.3(-/+)(0.1) were achieved for CIFAR10 and CIFAR100, respectively. We observed that the proposed method outperforms the commonly used pooling layers in CNN models. Code for this paper can be publicly accessed at: https://github.com/cuneytozdemir/lbppoolingÖğe Adaptive High Dynamic Range(IEEE, 2015) Atas, Musa; Dogan, YahyaInvestigating High Dynamic Range (HDR) approaches in the literature, a new and adaptive HDR model is developed in this study. HDR is processed based on images taken as a Low Dynamic Range (LDR) scheme that ranges between low exposure and high exposure values. Here, main focus is to present and to interpret challenging scenes or cases without having information loss by extending intensity ranges of a camera. Images, converting to HDR from LDR actually have a good satisfaction with respect to the information content, yet they are subject to effect which may deteriorate their natural quality. With respect to end user view, still it is hard to say HDR images satisfy photo-realistic characteristics. In this study, it is focused on information gain without detriment natural characteristics of the picture and thus a new HDR algorithm was developed. Proposed HDR method was crosschecked with famous methods used in the literature with regard to both photo-realistic picture quality, usability and computational cost criteria. It was observed that proposed method preferable over so called traditional algorithms and located in the bunch of first three methods by applying poll with 30 test subjects.Öğe Advancing brain tumor classification through MTAP model: an innovative approach in medical diagnostics(Springer Heidelberg, 2024) Ozdemir, Cuneyt; Dogan, YahyaThe early diagnosis of brain tumors is critical in the area of healthcare, owing to the potentially life-threatening repercussions unstable growths within the brain can pose to individuals. The accurate and early diagnosis of brain tumors enables prompt medical intervention. In this context, we have established a new model called MTAP to enable a highly accurate diagnosis of brain tumors. The MTAP model addresses dataset class imbalance by utilizing the ADASYN method, employs a network pruning technique to reduce unnecessary weights and nodes in the neural network, and incorporates Avg-TopK pooling method for enhanced feature extraction. The primary goal of our research is to enhance the accuracy of brain tumor type detection, a critical aspect of medical imaging and diagnostics. The MTAP model introduces a novel classification strategy for brain tumors, leveraging the strength of deep learning methods and novel model refinement techniques. Following comprehensive experimental studies and meticulous design, the MTAP model has achieved a state-of-the-art accuracy of 99.69%. Our findings indicate that the use of deep learning and innovative model refinement techniques shows promise in facilitating the early detection of brain tumors. Analysis of the model's heat map revealed a notable focus on regions encompassing the parietal and temporal lobes.Öğe Advancing early diagnosis of Alzheimer's disease with next-generation deep learning methods(Elsevier Sci Ltd, 2024) Ozdemir, Cuneyt; Dogan, YahyaAlzheimer's disease, characterized by cognitive decline and memory impairment, poses a significant healthcare challenge. This study presents a specially designed CNN model, utilizing contemporary approaches, to distinguish between various types of Alzheimer's disease. This model can serve as an early diagnostic tool to prevent the disease from progressing towards more pronounced and severe dementia symptoms. In this context, the performance of various transfer learning models has been examined, leading to the development of a specialized model integrating compression and excitation blocks, an innovative Avg-TopK pooling layer, and the SMOTE technique to handle data imbalance. The ablation study results demonstrate the critical role of these components, highlighting the model's effectiveness and innovative design. This study is novel in that it combines modern methodologies for detecting Alzheimer's disease, resulting in a model with state-of-the-art accuracy of 99.84% and improved computing efficiency. Grad-CAM analysis further demonstrates that the model focuses on cortical areas during classification, underscoring its potential as a robust diagnostic tool. These innovations represent a significant advancement over existing models, positioning this study as a pioneering effort in the early diagnosis of Alzheimer's disease. This study aims to contribute significantly to both academic research and medical applications by focusing on integrating artificial intelligence methodologies into medical diagnosis.Öğe An innovative approach for parkinson’s disease diagnosis using CNN, NCA, and SVM(Springer Science and Business Media Deutschland GmbH, 2024) Dogan, YahyaParkinson’s disease (PD) is a prevalent neurodegenerative disorder affecting millions of people globally, with substantial health risks and economic burdens. This study aims to introduce an innovative hybrid approach combining deep learning and machine learning algorithms to improve the diagnosis of PD using handwriting dynamics indicative of Parkinson’s symptoms. The proposed approach integrates hybrid feature extraction using nine fine-tuned transfer learning models, i.e., InceptionV3, DenseNet201, EfficientNetB0, ResNet50, MobileNetV2, VGG16, Xception, NASNetMobile, and InceptionResNetV2. Initially, features from these models are used individually or in binary and ternary combinations. Given the limited sample size in PD datasets, some extracted features through fine-tuning may lack significance, and fully connected layers can lead to overfitting. To address this issue, Neighborhood Component Analysis (NCA) is employed to refine these features, retaining only the most informative ones. Finally, the selected features are classified using Support Vector Machines (SVM) maximizing the margin between classes and reducing the risk of overfitting. The proposed hybrid model achieves a state-of-the-art accuracy of 99.39% on the Parkinson Hand Drawings dataset. The combination of features extracted from DenseNet201, Xception, and NASNetMobile models, processed using NCA and SVM methods, has been identified as the most efficient model, balancing high accuracy with computational efficiency. Qualitative assessments further confirm the accuracy and reliability of the approach. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.Öğe Chess Playing Robotic Arm(IEEE, 2014) Atas, Musa; Dogan, Yahya; Atas, IsaIn this study, a chess playing robotic arm system which has 5 degree of freedoms is developed. System comprised with various modules such as; main controller, image processing, machine learning, game engine and motion engine of robotic arm. Image processing unit is triggered only whenever opponent starts to move chessman. Meanwhile, images acquired in a specific time intervals are transmitted to the machine learning unit for classification purpose. After the classification process is taken place, opponent valid move is sent to the game engine as an input in order to generate reasonable output. Generated output is forwarded to the motion engine for positioning the robotic arm. It was observed that, developed system provides an efficient, favorable and immersive experience for player.Öğe Deep Learning Based Gender Identification Using Ear Images(Int Information & Engineering Technology Assoc, 2023) Kilic, Safak; Dogan, YahyaThe classification of an individual as male or female is a significant issue with several practical implications. In recent years, automatic gender identification has garnered considerable interest because of its potential applications in e-commerce and the accumulation of demographic data. Recent observations indicate that models based on deep learning have attained remarkable success in a variety of problem domains. In this study, our aim is to establish an end-to-end model that capitalizes on the strengths of competing convolutional neural network (CNN) and vision transformer (ViT) models. To accomplish this, we propose a novel approach that combines the MobileNetV2 model, which is recognized for having fewer parameters than other CNN models, with the ViT model. Through rigorous evaluations, we have compared our proposed model with other recent studies using the accuracy metric. Our model attained state-of-the-art performance with a remarkable score of 96.66% on the EarVN1.0 dataset, yielding impressive results. In addition, we provide t-SNE results that demonstrate our model's superior learning representation. Notably, the results show a more effective disentanglement of classes.Öğe Enhancing CNN model classification performance through RGB angle rotation method(Springer Science and Business Media Deutschland GmbH, 2024) Dogan, Yahya; Ozdemir, Cuneyt; Kaya, YılmazIn recent years, convolutional neural networks have significantly advanced the field of computer vision by automatically extracting features from image data. CNNs enable the modeling of complex and abstract image features using learnable filters, eliminating the need for manual feature extraction. However, combining feature maps obtained from CNNs with different approaches can lead to more complex and interpretable inferences, thereby enhancing model performance and generalizability. In this study, we propose a new method called RGB angle rotation to effectively obtain feature maps from RGB images. Our method rotates color channels at different angles and uses the angle information between channels to generate new feature maps. We then investigate the effects of integrating models trained with these feature maps into an ensemble architecture. Experimental results on the CIFAR-10 dataset show that using the proposed method in the ensemble model results in performance increases of 9.10 and 8.42% for the B and R channels, respectively, compared to the original model, while the effect of the G channel is very limited. For the CIFAR-100 dataset, the proposed method resulted in a 17.09% improvement in ensemble model performance for the R channel, a 5.06% increase for the B channel, and no significant improvement for the G channel compared to the original model. Additionally, we compared our method with traditional feature extraction methods like scale-invariant feature transform and local binary pattern and observed higher performance. In conclusion, it has been observed that the proposed RGB angle rotation method significantly impacts model performance. © The Author(s) 2024.Öğe Prediction of Adaptive Exposure Time in Hyperspectral Bands for Industrial Cameras(IEEE, 2015) Dogan, Yahya; Atas, MusaIn this study, a new method for exposure time correction for hyperspectral imaging is introduced. Initially, hardware setup was established. Then, a look-up table holds the minimum and maximum exposure times for each band was built. By using the developed image acquisition system, images having different exposure times for each hyperspectral band were acquired. After that, various features that can represent the exposure state were identified and a dataset was established. Prediction performance of the proposed method was cross validated by artificial neural network and outcomes were interpreted. It is observed that, by using the proposed method desired exposure quality can be determined with 99% accuracy.Öğe RGB-Angle-Wheel: A new data augmentation method for deep learning models(Elsevier, 2024) Ozdemir, Cuneyt; Dogan, Yahya; Kaya, YilmazDeep learning models often rely on a diverse and well -augmented dataset for optimal performance. In this context, the methods of data augmentation are pivotal in boosting the models' ability to generalize. In this paper, we introduce a novel data augmentation method, which we call RGB-Angle-Wheel, to improve the performance of deep learning models on RGB format images. This method involves rotating each color channel at specific angles to generate new training data that is distinct from the original dataset but shares similar properties. Experimental results on the CIFAR-10, CIFAR-100,and COCO datasets have validated the efficacy of the proposed method for enhancing model performance. Specifically, certain transformations in the red (R) and blue (B) channels improve model accuracy significantly, whereas the effect on the green (G) channel remains limited. These results indicate that the careful selection of transformation parameters plays a critical role in enhancing model performance. The findings of the study indicate that the proposed method can be utilized specifically for image processing, image classification, object detection, and other deep learning applications. Experiments demonstrate that the proposed method improves the model's efficacy and generalizability.
