DSpace Arşivi :: by Yazar "Talu, Muhammed Fatih" değerine göre listeleniyor

Yazar "Talu, Muhammed Fatih" seçeneğine göre listele

Listeleniyor 1 - 20 / 54

Artificial neural network regression model to predict flue gas temperature and emissions with the spectral norm of flame image
(Elsevier Sci Ltd, 2019) Golgiyaz, Sedat; Talu, Muhammed Fatih; Onat, Cem
This paper presents an experimental study on flue gas temperature (FGT) and emissions estimation in home-type nut coal-fired burner. The proposed method does not require prior knowledge of Charge-Coupled Device (CCD) camera features. Therefore, it can be applied easily without costly and complex adaptation requirement to control the combustion process. In the proposed system, the flame image was taken with a CCD camera. At the same time, reference temperature and emissions were taken with flue gas analyzer. Combustion characteristics were extracted by image processing techniques from each two-color channels of the flame image. When the features were obtained, instead of converting the flame image to grayscale and obtaining the general features, local feature extraction was preferred from each of the two-color channels that express the combustion process better. For this process, the image was divided into local windows and individual features for each two-color channel was extracted. The optimum number of windows was decided by experimental investigation. The features were obtained by using the spectral norm of the region of interest. The obtaining image features were used to train the Artificial Neural Network (ANN) regression model which predicted the FGT and emissions. Estimation accuracy (correlation coefficient (R)) of developed FGT prediction model is 0.99. The emission prediction models estimate SO2, O-2, NOx, CO2 andCO emissions with R = 0.97, R = 0.96, R = 0.77, R = 0.96, and R = 0.87 accuracies, respectively. The experimental results show that the FGT and emissions can be estimated by the flame image.
Atrial fibrillation classification and detection from ECG recordings
(Elsevier Sci Ltd, 2023) Gunduz, Ali Fatih; Talu, Muhammed Fatih
Objective: Atrial fibrillation (AF) heart rhythm disorder is investigated under two topics: Persistent AF (PeAF) and Paroxysmal AF (PAF). Diagnosis and detection of PeAF is relatively easier than PAF and PAF generally remains unrecognized. It is observed that a significant number of studies in the literature focused on detection of AF.Methods: In this study, four different approaches are examined for AF detection. The first one is based upon spectral features obtained from windowed ECG signals. In the second approach, distances between successor R peaks are used as features. Then in the third approach, P waves are detected from the ECG signals by using R peak positions and then the model is trained by those P waves. In those three approaches a deep learning ar-chitecture with bidirectional long short-term memory (BiLSTM) network is used. Finally, in the fourth approach, a convolutional long short-term memory (CLSTM) model with convolution and LSTM layers is used for classi-fication. The data set used in this work is obtained from 4th China Physiological Signal Challenge-2021.Results: As the result of experimental studies, it is seen that classification approach based on spectral features provided the best training accuracy (0.9788) and classification based on P wave detection provided the best test accuracy (0.8765).Significance: This study compares PeAF and PAF detection and classification methods based on deep learning models using different approaches. BiLSTM networks being capable of reflecting time sensitive features of ECG, appeared to be superior to CNN and LSTM cascades.
Aynalı Sazan (Cyprinus carpio) Balığında İmaj J-Fiji ile Sperm Hücresi Hareketlilik ve Hız Analizi
(2018) Erdoğan, Selim; Gürçay, Selahattin; Kocamaz, Adnan Fatih; Ateş, Burhan; Talu, Muhammed Fatih; Okumuş, Fatih; Özgür, Mustafa Erkan
Öz: Öz: Bu çalışma İmage J-Fiji programıyla sperm hücrelerine ait hareket değerlendirme yöntemi, bu yöntemin uygulanabilirliği, kolaylığı ve/veya zorluğunu araştırılması amacıyla yapılmıştır. Elde edilen parametreler, uluslararası yayınlarda sunulmuş benzer bilgisayar sistemleri ile elde edilmiş verilerle karşılaştırmaları yapılmıştır. Çalışmada aynalı sazan (Cyprinus carpio) türü balıklara ait sperm örnekleri incelenmiştir. İncelenen sperm hücrelerine ait hareketlilik parametreleri sırasıyla VSL: 74.05 ?m/sn, VCL: 115.18 ?m/sn, VAP: 63.85?m/sn, BCF: 10.75 Hz ve ALH: 19.28?m olarak hesaplanmıştır. Sonuç olarak, İmaj J-Fiji programı ile balıklarda sperm hücresi yakalama, işleme ve değerlendirme işlemlerinin kolay, uygulanabilir olduğu ve balık üretim merkezlerinde erkek damızlık balıkların sperm kalitesinin belirlenmesinde pratik ve hızlı bir yöntem sunduğu söylenebilir.
Boyut İndirgeme Yöntemlerinin Karşılaştırmalı Analizi
(2020) Talu, Muhammed Fatih; Çalışan, Mücahit
Günümüz veritabanları hızlı bir şekilde büyümektedir. Örneğin Youtube’a her dakikada ortalama 300 saatlik video yüklenmektedir. Veri boyutuyla orantılı bir şekilde, işleme, depolama ve transfer maliyetleri artmaktadır. Buna karşılık, özellikle video ve imge gibi yüksek boyutlu veri içeriklerinin büyük oranda benzer olduğu bilinmektedir. Bu tür yüksek boyutlu ham verilerin, düşük boyutlara indirgenmesi, imge sınıflandırma, algılama ve anlamlı bilgi çıkarım prosesleri için hayati öneme sahiptir. Veri boyutunu indirgeyen çok sayıda teknik mevcuttur. Klasik yapay öğrenme tekniklerinden; PCA (Temel Bileşenler Analizi) ve LDA (Doğrusal Ayıraç Analizi), probleme matematiksel bir çözüm zemini kazandırdıkları için ön plana çıkarken, doğrusal olmayan tekniklerden, derin öğrenme yaklaşımlarından olan Oto-Kodlayıcı (Auto-Encoding), büyük verilerin indirgenmesine izin vermesi bakımından araştırmacıların ilgisini çekmektedir. Bu çalışmada, gerçek ve sentetik veriler (doğrusal ve doğrusal olmayan) kullanılarak PCA, LDA ve Auto-Encoding (AE) yöntemlerinin boyut indirgeme performansları incelenmiştir. Belirli kıstaslarda (harcanan zaman, yeniden inşa etme doğruluğu vb.) alınan sonuçlar karşılaştırmalı bir şekilde sunulmuştur.
Brain MRI high resolution image creation and segmentation with the new GAN method
(Elsevier Sci Ltd, 2023) Guven, Sara Altun; Talu, Muhammed Fatih
Brain magnetic resonance imaging segmentation is a recent and still popular research area. Good and accurate segmentation results play an important role in the diagnosis of cancer or other brain diseases. In this article, a novel Generative Adversarial Network architecture is proposed for brain magnetic resonance imaging segmentation. The proposed method is called SSimDCL (Supervised SimDCL). Four studies were conducted in this article. In the first study, the SSimDCL method on the two-dimensional brain magnetic resonance imaging dataset was compared with the current state-of-art architectures CycleGAN, CUT, FastCUT, DCLGAN, and SimDCL. In the second study, the dataset resolution was improved. In the third study, being measured the efficiency of the newly created dataset. And the SSimDCL is trained for both the dataset with increased resolution and the normal dataset, and the results are obtained. In the fourth study, the results of the SSimDCL and the VolBrain brain magnetic resonance imaging segmentation results, which are widely used today, are included. When VolBrain segmentation and SSimDCL segmentation are compared. The results were compared both visually and metrically. Fre ' chet Inception Distance (FID), Kernel Inception Distance (KID), Peak Signal to Noise Ratio (PSNR) and Learned Perceptual Image Patch Similarity (LPIPS) were used as measurement metrics. The Jaccard and Dice similarity metrics were also used in the analysis. It was observed that the SSimDCL give satisfactory results in all four studies. This method can be used as an automatic brain MRI image segmentation system.
Bıyık Deseni Üretiminde Çekişmeli Üretici Ağların Performans Karşılaştırması
(2021) Şahin, Emrullah; Talu, Muhammed Fatih
Bu çalışmada görüntüden görüntüye dönüşüm yapan çekişmeli üretici ağ mimarilerinin performans incelemesi yapılıp, sentetik görüntü üretimindeki başarımı değerlendirilmiştir. Bu modellerin kaliteli bir başarım değerlendirmesi için standartlaştırılmış veri kümeleri yerine gerçek iş alanından toplanılan denim2bıyık veri kümesi kullanılmıştır. Denim kumaşları üzerine çizilen bıyık desenleri lazer cihazıyla oluşturulmaktadır. Bu cihazın istenilen bıyık desenini oluşturabilmesi için uzmanlaşmış bir personel tarafından görsel düzenleme programları ile yaklaşık 2-3 saat süren bir çalışma yapması gerekir. Önerilen yaklaşımla otomatik bir bıyık üretim işlemi gerçekleşecek, manuel üretimdeki hatalar ve zamansal kayıplar elimine edilecektir. Yaptığımız literatür araştırması neticesinde denim ürün görsellerinin üretken ağlar ile üretilmesi ile ilgili farklı bir çalışma bulunmamaktadır. Bu durum yapılan çalışmanın akademik özgün değerini yükseltmektedir. Çalışmada kullanılan ÇÜA mimarileri Pix2Pix, CycleGAN, DiscoGAN ve AttentionGAN’dır. Her bir mimarinin denim2bıyık veri kümesindeki eğitim ve test verileri üzerinde bıyık deseni üretim başarım değerlendirmesi ve maliyet analizi yapılmıştır. Yapılan çalışmalar sonucunda, bıyık desen görseli üretim hızı bir saniyenin altına düşerken, üretim doğruluğu %86 seviyelerine çıktığı görülmektedir.
Cat Swarm Based Shadow Detection Method
(Ieee, 2018) Firildak, Kazim; Karaduman, Mucahit; Talu, Muhammed Fatih; Yeroglu, Celaleddin
Innovations brought by developing technology are used in different forms in security area. Those technologies on imaging are usually involved in monitoring, tracking and detecting. When these processes are performed, the shadow of the object can prevent detection and monitoring. Therefore, in this study, a method has been proposed to determine the shadows of the objects and to distinguish them from the original image, and to determine the real image of the object. Previous work on this area has been done using classic shadow detection methods. In order to overcome their disadvantages during the detection period, this study uses cat swarm optimization, which can be applied to many problems. With this method, the shadows belonging to the object are detected, and the objects are determined by separating the shadows from the image. Later, if desired, follow-up of the object will also be achieved. It is shown that the suggested cat swarm optimization algorithm provides an effective result in shadow detection and give better results with shadow detection rate over compared algorithms.
Çekişmeli üretken ağ modellerinin görüntü üretme performanslarının incelenmesi
(2020) Çelik, Gaffari; Talu, Muhammed Fatih
Derin öğrenme alanında yaşanan en önemli gelişmelerden biri, hiç şüphesiz çekişmeli üretken ağ (Generative adversarial network-GAN) modelleridir. GAN olarak anılan bu modeller, görüntü veri kümesinin genişletilmesinde (image augmentation), resim/karikatür boyamada (painting), yüksek çözünürlüğe sahip süper görüntü elde etmede, bir görüntüdeki doku/desenin başka bir görüntüye transferinde kullanılan en modern yaklaşımlar olarak karşımıza çıkmaktadır. Bu çalışmada literatürde yaygın olarak kullanılan GAN modellerinin (cGAN, DCGAN, InfoGAN, SGAN, ACGAN, WGAN-GP, LSGAN), gerçek görüntülere çok benzeyen sentetik görüntüleri üretmedeki performansları incelenmiştir. Çalışmanın orijinalliği, cGAN ve DCGAN’ın avantajlarını barındıran hibrit bir GAN modeli (cDCGAN) geliştirilmesi ve GAN yöntemlerinin performansları, derin öğrenme tabanlı evrişimsel sinir ağları(CNN) ile kıyaslamalı olarak değerlendirmesidir. Kodlanan modellerle veri kümelerindeki görüntülere benzer sentetik görüntüler üretilmiştir. Üretilen sentetik görüntülerin mevcut görüntülere benzerliklerini hesaplamak, böylece model performansını değerlendirebilmek için fréchet başlangıç mesafesi (FID) metriği ve CNN kullanılmıştır. Yapılan deneysel çalışmalarda, tüm modellerin zamana bağlı görüntü üretim performansları değerlendirilmiştir. Sonuç olarak, LSGAN modeliyle üretilen görüntülerin yüksek sınıflandırma başarım oranı sağladığı, ancak DCGAN ve WGANGP ile daha gürültüsüz net görüntüler ürettiği gözlemlenmiştir.
CNN-Based Fabric Defect Detection System on Loom Fabric Inspection
(E.U. Printing And Publishing House, 2022) Talu, Muhammed Fatih; Hanbay, Kazim; Varjovi, Mahdi Hatami
Fabric defect detection is generally performed based on human visual inspection. This method is not effective and it has various difficulties such as eye delusion and labor cost. To deal with these problems, machine learning and computer vision-based intelligent systems have been developed. In this paper, a novel real-time fabric defect detection system is proposed. The proposed industrial vision system has been operated in real-time on a loom. Firstly, two fabric databases are constructed using real fabric images and new defective patch capture (DPC) algorithm. One of the main objectives in this study is to develop a CNN architecture that focuses only on fabric defect detection. One of the most unique aspects of the study is to detect defective pixel regions of fabric images with Fourier analysis on a patch-based and integrate it with deep learning Thanks to the novel developed fast Fourier transform-based DPC algorithm, defective texture areas become visible and defect-free areas are suppressed, even on complex denim fabric textures. Secondly, an appropriate convolution neural networks (CNN) model is developed. Thus the new dataset dataset is refined using negative mining method and CNN model. However, traditional feature extraction and classification approaches are also used to compare classification performances of deep models and traditional models. Experimental results show that our proposed CNN model integrated with negative mining can classify the defected images with high accuracy. Also, the proposed CNN model has been tested in real-time on a loom, and it achieves 96.5% detection accuracy. The proposed model obtains better accuracy and speed performance in terms of detection accuracy with a much smaller model size.
Comparison of Extended-Kalman- and Particle-Filter-Based Sensorless Speed Control
(Ieee-Inst Electrical Electronics Engineers Inc, 2012) Aydogmus, Omur; Talu, Muhammed Fatih
State estimation process is one of the major concerns for controlling and monitoring systems in industry which requires high-cost measurements or unmeasurable variables of nonlinear systems. These drawbacks can be highly eliminated by designing systems without using any kind of sensors. In this paper, sensorless speed control of a dc motor was performed by using extended Kalman filter (EKF) and particle filter (PF). The speed information is estimated by using armature current data measured from a dc motor which is controlled in various speed references with a closed-loop controller. Furthermore, a performance comparison of the EKF and the PF by taking into consideration their estimation errors under the same conditions was realized in a simulation environment. The comparison results showed that the estimation performance of the PF is more accurate but slower than the EKF. The quantitative values of accurateness and slowness are depended on the particle number of the PF. The obtained computation times of the PF having ten particles and the EKF are 180 and 15 mu s, respectively.
Comparison of methods for determining activity from physical movements
(2021) Çalışkan, Mücahit; Talu, Muhammed Fatih
Abstract: In this study, the methods which can detect the basic physical movements of a person (downward, upward, sitting, stop, walking,running) from inertial sensor (IMU) data are evaluated. The performances of classical (ANN, SVM, k-NN) and current approaches(Convolutional Neural Networks-ESA) to map IMU data to activity classes were compared. A three-stage study was carried outfor this aim: 1) data acquisition; 2) creating training/test sets; 3) construction and classification of network architectures. At thestage of data acquisition, to obtain 6 different physical movements from 10 different people, the accelerometer sensor is placed onthe persons. Repetitive movements of persons were recorded. At the second stage, the recorded long-term accelerometer data isdivided into packages in the form of short-term windows. The training set of classical approaches was constructed by featuresextracting from each packet data containing one-dimensional acceleration information. The transformation of one-dimensionalsignals to a two-dimensional image matrix for the training set of the deep learning-based approaches was performed. In the thirdstage, ANN, SVM, k-NN and CNN architectures were constructed, and classification process was carried out. As a result of theexperimental studies, it was found that the accuracy of IMU-activity mapping was 99% with the ANN method and 95% with theCNN method.
Continuous rotation invariant features for gradient-based texture classification
(Academic Press Inc Elsevier Science, 2015) Hanbay, Kazim; Alpaslan, Nuh; Talu, Muhammed Fatih; Hanbay, Davut; Karci, Ali; Kocamaz, Adnan Fatih
Extracting rotation invariant features is a valuable technique for the effective classification of rotation invariant texture. The Histograms of Oriented Gradients (HOG) algorithm has been proved to be theoretically simple, and has been applied in many areas. Also, the co-occurrence HOG (CoHOG) algorithm provides a unified description including both statistical and differential properties of a texture patch. However, HOG and CoHOG have some shortcomings: they discard some important texture information and are not invariant to rotation. In this paper, based on the original HOG and CoHOG algorithms, four novel feature extraction methods are proposed. The first method uses Gaussian derivative filters named GDF-HOG. The second and the third methods use eigenvalues of the Hessian matrix named Eig(Hess)-HOG and Eig(Hess)-CoHOG, respectively. The fourth method exploits the Gaussian and means curvatures to calculate curvatures of the image surface named GM-CoHOG. We have empirically shown that the proposed novel extended HOG and CoHOG methods provide useful information for rotation invariance. The classification results are compared with original HOG and CoHOG algorithms methods on the CUReT, KTH-TIPS, KTH-TIPS2-a and UIUC datasets show that proposed four methods achieve best classification result on all datasets. In addition, we make a comparison with several well-known descriptors. The experiments of rotation invariant analysis are carried out on the Brodatz dataset, and promising results are obtained from those experiments. (C) 2014 Elsevier Inc. All rights reserved.
Derin Sinir Ağları için Hiperparametre Metodlarının ve Kitlerinin İncelenmesi
(2021) Altun, Sara; Talu, Muhammed Fatih
Otomatik makine öğrenimi (AutoML) ve derin sinir ağları birçok hiperparametreye sahiptir. Karmaşık ve hesapsal maliyet olarak pahalı makine öğrenme modellerine son zamanlarda ilginin artması, hiperparametre optimizasyonu (HPO) araştırmalarının yeniden canlanmasına neden olmuştur. HPO’nun başlangıcı epey uzun yıllara dayanmaktadır ve derin öğrenme ağları ile popülaritesi artmıştır. Bu makale, HPO ile ilgili en önemli konuların gözden geçirilmesini sağlamaktadır. İlk olarak model eğitimi ve yapısı ile ilgili temel hiperparametreler tanıtılmakta ve değer aralığı için önemleri ve yöntemleri tartışılmaktadır. Sonrasında, özellikle derin öğrenme ağları için etkinliklerini ve doğruluklarını kapsayan optimizasyon algoritmalarına ve uygulanabilirliklerine odaklanılmaktadır. Aynı zamanda bu çalışmada HPO için önemli olan ve araştırmacılar tarafından tercih edilen HPO kitlerini incelenmiştir. İncelenen HPO kitlerinin en gelişmiş arama algoritmaları, büyük derin öğrenme araçları ile fizibilite ve kullanıcılar tarafından tasarlanan yeni modüller için genişletilebilme durumlarını karşılaştırmaktadır. HPO derin öğrenme algoritmalarına uygulandığında ortaya çıkan problemler, optimizasyon algoritmaları arasında bir karşılaştırma ve sınırlı hesaplama kaynaklarına sahip model değerlendirmesi için öne çıkan yaklaşımlarla sonuçlanmaktadır.
EEG sinyallerinden bakılan görselin üretilmesi
(2021) Talu, Muhammed Fatih; Çelik, Gaffari
Öz: EEG sinyalleri kullanılarak engelliler için kontrol edilebilir tekerlekli sandalyelerin üretildiği veya yapılması düşünülen aktivitenin tahmin edildiği çalışmalara literatürde sıklıkla rastlanmaktadır. Genel olarak bu çalışmalarda elektroensefalografi (EEG) sinyalinin önceden belirlenen sınıflara aktarımı gerçekleştirilir. Bu çalışmalar EEG sinyalinin sınıflandırmasından ibarettir. Ancak son yıllarda yapay öğrenme alanında yaşanan gelişmelerle sınıflandırmadan öteye gidildiği, EEG sinyalinden bakılan görselin üretilebildiği görülmektedir. Klasik çekişmeli üretici ağlar (Generative adversarial networks-GAN) ve otomatik kodlayıcı (Auto encoder-AE) yaklaşımlarının kullanıldığı sınırlı sayıdaki bu çalışmalar incelendiğinde, EEG sinyallerinden kabaca görsellerin üretilebildiği görülmektedir. Bu çalışmanın özgün yönü, görsel üretim kabiliyetini arttıracak matematiksel yaklaşımlar içermesidir. Klasik GAN mimarileri üretilen görüntülerin çeşitliliğini sağlayabilmek için rastgele vektör girişini kullanırlar. Bu yaklaşım ile EEG sinyalinden üretilen görsellerin düşük kalitede olduğu gözlemlenmiştir. Önerilen yöntemde giriş iki kısım (kodlanmış EEG ve rastgelelik) olarak düşünülmüştür. EEG’nin kodlanması için değişken oto kodlayıcı (Variational auto encoder-VAE) ve fourier dönüşümü (FD) kullanılırken, rastgelelik için iki farklı yaklaşım önerilmiştir. Bu özgün GAN kullanımı, EEG sinyallerinden daha kaliteli görsel üretilmesini sağlamıştır. Bu kalitenin sayısal olarak anlaşılabilmesi için önceden eğitilmiş evrişimsel sinir ağları (ESA) kullanılmıştır. Yapılan deneysel çalışmalar neticesinde, klasik GAN ile EEG’den üretilen görsellerin başarım seviyesi %93 civarındayken, önerilen yaklaşımda bu seviyenin %95-%100 aralığına çıktığı görülmektedir.
Estimation of excess air coefficient on coal combustion processes via gauss model and artificial neural network
(Elsevier, 2022) Golgiyaz, Sedat; Talu, Muhammed Fatih; Das, Mahmut; Onat, Cem
It is no doubt that the most important contributing cause of global efficiency of coal fired thermal systems is combustion efficiency. In this study, the relationship between the flame image obtained by a CCD camera and the excess air coefficient (lambda) has been modelled. The model has been obtained with a three-stage approach: 1) Data collection and synchronization: Obtaining the flame images by means of a CCD camera mounted on a 10 cm diameter observation port, lambda data has been coordinately measured and recorded by the flue gas analyzer. 2) Feature extraction: Gridding the flame image, it is divided into small pieces. The uniformity of each piece to the optimal flame image has been calculated by means of modelling with single and multivariable Gaussian, calculating of color probabilities and Gauss mixture approach. 3) Matching and testing: A multilayer artificial neural network (ANN) has been used for the matching of feature-lambda. (C) 2021 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Alexandria University.
Estimation of Upper Extremity Movement Performance in Stroke Patients with Artificial Learning Techniques
(2021) Çalışan, Mücahit; Talu, Muhammed Fatih
Abstract: The main reason why people are directed to rehabilitation services after stroke-like neurological diseases are to bring individuals' daily abilities to a normal level. Measuring the activities of people in their daily lives ensures that these rehabilitation services progress more healthily. In our study, Brunnstrom Hemiplegia Recovery Staging, which is widely used by doctors to evaluate the movement function of stroke patients during rehabilitation, was examined. The study was specifically adapted to the upper extremity stage 4a movement of the Brunnstrom Staging. Daily movements of patients were evaluated with accelerometer sensors. With this methodology, sensor data was collected from 15 volunteer stroke patients and 80 healthy individuals. These sensor data were interpreted by the medical professional. Thus, consistency between movement data of healthy and sick individuals was analyzed. The data obtained as a result of the analysis process were examined with artificial learning methods and classified as healthy/unhealthy. The methodology of the study is suitable for research designed to increase upper / lower extremity performance in the daily life of individuals.
Extended kalman filter based IMU sensor fusion application for leakage position detection in water pipelines
(Gazi Univ, Fac Engineering Architecture, 2017) Akkaya, Abdullah Erhan; Talu, Muhammed Fatih
In water distribution networks, there is a serious loss of water due to cracks and faults in pipes. It is very important to detect these failures and fix them in a short period of time to prevent the loss of water and related income. Instead of general repair operations with high costs in distribution networks, finding the exact location of the fault and only working in that area will reduce the repair costs. Although superficial pipe listening devices seem to be the solution to this need, it is not preferred because of affected by the ambient sounds that reducing the efficiency of this method. GPS-based leak detection systems, which are commercially available on the market, can operate on large-scale water pipelines (>= 6 inch) and have high costs. In this study, we present a preliminary study of a leak detection robot prototype that can operate on smaller diameter pipes and without the need for a GPS system. In this study, a preliminary study of a leak detection robot prototype that can operate on smaller diameter pipes and without the need for a GPS system was presented. In this preliminary study, the design, production, location and leakage prediction software of a robot can move with the pushing force of water in the pipeline has been realized. The position estimation is performed by using the 9-DOF IMU (3D-accelerometer, 3D-gyroscope and 3D-magnetometer) sensor data in the Extended Kalman Filter. The leakage estimation includes the location of the corresponding peak point in the instantaneous recorded sound data. In the performed experimental studies, it was seen that the leakage location estimation error in the total 118m navigation result is about 0.25m.
Fabric Defect Detection Methods for Circular Knitting Machines
(Ieee, 2015) Hanbay, Kazim; Talu, Muhammed Fatih; Ozguven, Omer Faruk; Ozturk, Dursun
In this paper, an online fabric defect detection system that can detect fabric defects which may occur during the fabric product in knitting machines is introduced. This system mainly includes three steps: 1) Construction of a defected/defect-free fabric database; 2) Obtaining and classification of the feature vectors; 3) Online working on embedded system. This study only contains information about the first two stages. In the first stage, 3242 'defected' and '5923' defect-free images were acquired by using a conveyor system which has line scan camera and linear light. In the second stage, filtering, feature extraction (wavelet transform, co-occurrence matrix and CoHOG) and classification (YSA) processes were carried out. As a result, obtaining the feature vectors through wavelet transform has reduced computation cost by 53% and also has successfully provided the classification of the defects by 90%.
Fabric defect detection systems and methods-A systematic literature review
(Elsevier Gmbh, 2016) Hanbay, Kazim; Talu, Muhammed Fatih; Ozguven, Omer Faruk
This paper presents a comprehensive literature review of fabric defect detection methods First, it briefly explains basic image acquisition system components such as camera and lens. Defect detection methods are categorized into seven classes as structural, statistical, spectral, model-based, learning, hybrid and comparison studies. These methods are evaluated according to such criteria as the accuracy, the computational cost, reliability, rotating/scaling invariant, online/offline ability to operate and noise sensitivity. Strengths and weaknesses of each approach are comparatively highlighted. In addition, the availability of utilizing methods for weaving and knitting in machines is investigated. The available review studies do not provide sufficient information about fabric defect detection systems for readers engaged in research in the area of textile and computer vision. A set of examination for efficient establishment of image acquisition system are added. In particular, lens and light source selection are mathematically expressed. (C) 2016 Elsevier GmbH. All rights reserved.
Fabric Defect Detection Using Customized Deep Convolutional Neural Network for Circular Knitting Fabrics
(2022) Varjovi, Mahdi Hatami; Talu, Muhammed Fatih; Hanbay, Kazım
Visual inspection is a main stage of quality assurance process in many applications. In this paper, we propose a new network architecture for detecting the fabric defects based on convolutional neural network. Four different pre-trained and customized model network architectures have compared in terms of performance. Results has been evaluated on a fabric defect dataset of 13.800 images. Among the existing Inception V3, MobileNetV2, Xception and ResNet50 methods, the InceptionV3 model has achieved 78% classification success. Our designed deep network model could achieve 97% success. The experimental works show that the designed deep model is effective in detecting the fabric defects.

Yazar "Talu, Muhammed Fatih" seçeneğine göre listele

Sayfa Başına Sonuç

Sıralama seçenekleri