91 Hits in 1.2 sec

Quadruplet Selection Methods for Deep Embedding Learning [article]

Kaan Karaman, Erhan Gundogdu, Aykut Koc, A. Aydin Alatan
2019 arXiv   pre-print
For training a model that can extract proper features, the distance between two samples of a dataset in the feature space † This work was done when Erhan Gundogdu was with Middle East Technical University  ... 
arXiv:1907.09245v1 fatcat:hm6zm2ul2bderpbddwddcsqrqi

Shape Reconstruction by Learning Differentiable Surface Representations [article]

Jan Bednarik, Shaifali Parashar, Erhan Gundogdu, Mathieu Salzmann, Pascal Fua
2019 arXiv   pre-print
Generative models that produce point clouds have emerged as a powerful tool to represent 3D surfaces, and the best current ones rely on learning an ensemble of parametric representations. Unfortunately, they offer no control over the deformations of the surface patches that form the ensemble and thus fail to prevent them from either overlapping or collapsing into single points or lines. As a consequence, computing shape properties such as surface normals and curvatures becomes difficult and
more » ... liable. In this paper, we show that we can exploit the inherent differentiability of deep networks to leverage differential surface properties during training so as to prevent patch collapse and strongly reduce patch overlap. Furthermore, this lets us reliably compute quantities such as surface normals and curvatures. We will demonstrate on several tasks that this yields more accurate surface reconstructions than the state-of-the-art methods in terms of normals estimation and amount of collapsed and overlapped patches.
arXiv:1911.11227v1 fatcat:2ubd2s3a2bhn7nay7b5f2z7pfa

Contrastive Language-Action Pre-training for Temporal Localization [article]

Mengmeng Xu, Erhan Gundogdu, Maksim Lapin, Bernard Ghanem, Michael Donoser, Loris Bazzani
2022 arXiv   pre-print
Long-form video understanding requires designing approaches that are able to temporally localize activities or language. End-to-end training for such tasks is limited by the compute device memory constraints and lack of temporal annotations at large-scale. These limitations can be addressed by pre-training on large datasets of temporally trimmed videos supervised by class annotations. Once the video encoder is pre-trained, it is common practice to freeze it during fine-tuning. Therefore, the
more » ... eo encoder does not learn temporal boundaries and unseen classes, causing a domain gap with respect to the downstream tasks. Moreover, using temporally trimmed videos prevents to capture the relations between different action categories and the background context in a video clip which results in limited generalization capacity. To address these limitations, we propose a novel post-pre-training approach without freezing the video encoder which leverages language. We introduce a masked contrastive learning loss to capture visio-linguistic relations between activities, background video clips and language in the form of captions. Our experiments show that the proposed approach improves the state-of-the-art on temporal action localization, few-shot temporal action localization, and video language grounding tasks.
arXiv:2204.12293v1 fatcat:klsv3gmr7zgpdnud2ec7mhi5eu

Epidermal cyst mimicking incision line metastasis

Ramazan Gundogdu, Erhan Ayhan, Tahsin Colak
2017 Turkish Journal of Surgery  
Epidermal cysts are cystic tumors lined with keratinized squamous layer and filled with keratin debris. Epidermal cysts may develop by implantation of surface epidermal layer into the dermis or subcutaneous tissue after trauma or surgical procedures. Cervix cancer spreads either directly or via the vascular and lymphatic systems. Distant skin metastasis of endometrium or cervix cancer is very rare. In this case report, a patient who had a history of cervix cancer operation 11 years ago and
more » ... nted with a mass that mimicked incision line metastasis and was histopathologically diagnosed with epidermal cyst is presented.
doi:10.5152/ucd.2015.2927 pmid:28740968 pmcid:PMC5508240 fatcat:nxdk46tanfczjgwnxmtepbpgwu

Revamping Cross-Modal Recipe Retrieval with Hierarchical Transformers and Self-supervised Learning [article]

Amaia Salvador, Erhan Gundogdu, Loris Bazzani, Michael Donoser
2021 arXiv   pre-print
Cross-modal recipe retrieval has recently gained substantial attention due to the importance of food in people's lives, as well as the availability of vast amounts of digital cooking recipes and food images to train machine learning models. In this work, we revisit existing approaches for cross-modal recipe retrieval and propose a simplified end-to-end model based on well established and high performing encoders for text and images. We introduce a hierarchical recipe Transformer which
more » ... y encodes individual recipe components (titles, ingredients and instructions). Further, we propose a self-supervised loss function computed on top of pairs of individual recipe components, which is able to leverage semantic relationships within recipes, and enables training using both image-recipe and recipe-only samples. We conduct a thorough analysis and ablation studies to validate our design choices. As a result, our proposed method achieves state-of-the-art performance in the cross-modal recipe retrieval task on the Recipe1M dataset. We make code and models publicly available.
arXiv:2103.13061v1 fatcat:smg4gd3hevgxtgg2f6swyvlt3a

GarNet++: Improving Fast and Accurate Static3D Cloth Draping by Curvature Loss [article]

Erhan Gundogdu, Victor Constantin, Shaifali Parashar, Amrollah Seifoddini, Minh Dang, Mathieu Salzmann, Pascal Fua
2020 arXiv   pre-print
Gundogdu In the more sophisticated implementation depicted by Fig. 2b , we explicitly take them into account to further model the skin-cloth interactions.  ...  Gundogdu received the Ph.D. degree in Electrical and Electronics Engineering Dept. from Middle East Technical University (METU), Ankara, Turkey in 2017.  ... 
arXiv:2007.10867v1 fatcat:nbx4sfgcqnembjq5ywiwrvr5pu

Primary intracranial germinomas: retrospective analysis of five consecutively operated cases

Anas Abdallah, Murad Asilturk, Mustafa Levent Uysal, Erhan Emel, Betul Guler Abdallah, Bilge Bilgic, Gokcen Gundogdu
2016 Turkish Neurosurgery  
doi:10.5137/1019-5149.jtn.17474-16.3 pmid:27651341 fatcat:stip37pdfrg5heyjo6puknnm5y

Generic and attribute-specific deep representations for maritime vessels

Berkan Solmaz, Erhan Gundogdu, Veysel Yucesoy, Aykut Koc
2017 IPSJ Transactions on Computer Vision and Applications  
Fine-grained visual categorization has recently received great attention as the volumes of labeled datasets for classification of specific objects, such as cars, bird species, and air-crafts, have been increasing. The availability of large datasets led to significant performance improvements in several vision-based classification tasks. Visual classification of maritime vessels is another important task, assisting naval security and surveillance applications. We introduced, MARVEL, a
more » ... image dataset for maritime vessels, consisting of 2 million user-uploaded images and their various attributes, including vessel identity, type, category, year built, length, and tonnage, collected from a community website. The images were categorized into vessel type classes and also into superclasses defined by combining semantically similar classes, following a semi-automatic clustering scheme. For the analysis of the presented dataset, extensive experiments have been performed, involving several potentially useful applications: vessel type classification, identity verification, retrieval, and identity recognition with and without prior vessel type knowledge. Furthermore, we attempted interesting problems of visual marine surveillance such as predicting and classifying maritime vessel attributes such as length, summer deadweight, draught, and gross tonnage by solely interpreting the visual content in the wild, where no additional cues such as scale, orientation, or location are provided. By utilizing generic and attribute-specific deep representations for maritime vessels, we obtained promising results for the aforementioned applications.
doi:10.1186/s41074-017-0033-4 fatcat:juiehnoxqzdy3c72gcnm727r5a

Rheumatoid Arthritis: Are psychological factors effective in disease flare?

Volkan Yilmaz, Ebru Umay, Ibrahim Gundogdu, Zeliha Ozgur Karaahmet, Arif Erhan Ozturk
2017 European Journal of Rheumatology  
Rheumatoid arthritis (RA) is clinically an undulant disease, and reasons for flare or remission vary. We aimed to identify factors potentially associated with disease flare and remission. Two hundred and seventy-four patients with RA who were admitted to our center between January 2010 and January 2016 were included. Disease activity was evaluated using disease activity score 28 (DAS 28); functional status was evaluated using the modified Health Assessment Questionnaire (m-HAQ), a questionnaire
more » ... that comprises flare or remission domains such as psychological stress and mood status, physical trauma, nutrition regimen, infection, antibiotic use, and seasonal weather changes. Beck Depression Inventory (BDI) and Beck Anxiety Inventory (BAI) were used to identify if patients had a mood disorder. Four subscales of Arthritis Impact Measurement Scale-2 (AIMS-2) (level of tension, mood, general perception of health, and satisfaction with health) were used to determine patient quality of life. Of the 274 patients, 261 were female (95.3%) and 13 were male (4.7%); the mean age was 52.10±9.41 years. According to patients' perception, the most frequent reasons for joint symptoms were psychological stress/mood disorder (86.1%), followed by infection (49.6%) and trauma (46.4%). The most frequent factors for remission of symptoms were antibiotic use (42.7%), cold weather (34.3%), and hot weather (19%). Psychological stress and mood status are independent factors for relapse periods in patients with RA. These should be considered particularly in patients who are resistant to different treatment regimens and in whom any other reason for disease flare is not obvious.
doi:10.5152/eurjrheum.2017.16100 pmid:28638686 pmcid:PMC5473448 fatcat:dcebf62zpjdtpnjfkoltkphmjm

Radiation-Induced Benign Meningioma after Radiotherapy for Medulloblastoma: Case report and brief review of the literature

Hasan Burak Gündüz, Anas Abdallah, Erhan Emel, Gökçen Gündoğdu
2018 Sinir Sistemi Cerrahisi Dergisi  
Medulloblastoma is one of the most common primary malignant brain tumors of central nervous system in children. The recommended treatment of this malignant neoplasm is aggressive surgery, followed by craniospinal radiotherapy (RTP) with/without chemotherapy. Radiation-induced meningioma is the most common brain neoplasm known to be caused by ionizing radiation. This report presents a patient who had undergonde an operation 25 years ago because of a cerebral mass lesion. The patient had
more » ... ed unsteady gait associated with nausea and vomiting 25 years ago. Cranial CT demonstrated a large, solid and contrast-enhanced midline vermian mass. Through midline suboccipital craniectomy mass was removed totally which was histopathologically diagnosed as desmoplastic medulloblastoma. The patient was given postoperative adjuvant RTP. The patient tolerated RTP well and she had been called for yearly controls for 25.5 years. The patient who had not any complaint at routine controls for 25.5 years had been brought to our hospital with headache, unsteady gait, nausea and vomiting. MRI demonstrated an isointense mass in the right cerebellopontine angle that consisted of two compartments; the medial compartment did not show contrast enhancement and the lateral compartment that was associated with dura showed moderate, and homogenous contrast enhancement The patient had been managed surgically using right suboccipital craniectomy followed by total resection of the lateral compartment and biopsy of the medial compartment was obtained which was thought to be radiation-induced fibrotic tissue. The mass was diagnosed histopathologically as fibroblastic meningioma WHO Grade I and the biopsy of medial compartment showed to be arachnoidal fibrosis. The patient was well and had no recurrence öat postoperative 30 months. Keywords: Radiation-induced meningioma, medulloblastoma in children, late effect of radiation therapy J Nervous Sys Surgery 2016;6(1-2):61-66 Posterior Fossa Medülloblastomun Tedavisinde Kullanılan Kraniyal Radyasyon Geç Komplikasyonu olan Serebellopontin Köşenin Menenjiomu: olgu Sunumu ve Literatür Taraması Medülloblastom çocukluk çağında primer merkezi sinir sistem tümörlerinin en sık rastlanan malign beyin tümörlerinden biridir. Bu malign tümörün tedavisinde agresif cerrahi sonrasında kraniyospinal radyoterapi (RTP) ve/veya kemoterapi önerilmektedir. Radyasyonla ilişkili menenjiom, iyonize radyasyonun sonucu olarak en sık rastlanan beyin tümörüdür. Bu makalede, beyin kitlesi nedeniyle 25 yıl önce ameliyat geçiren hasta sunulmuştur. Yirmi beş yıl önce dengesiz yürüyüşle birlikte bulantı ve kusması vardı. Kraniyal BT'sinde büyük, sert, kontrast tutan orta hat vermiyan kitle izlenmiştir. Orta suboksipital kraniyektomi yapılarak histopatolojik olarak desmoplastik medülloblastom tanısı konulan kitle total olarak çıkartılmıştır. Hastaya ek olarak postoperatif RTP verildi. Hasta RTP'yi iyi tolere edilip 25,5 yıldır yıllık kontrollere çağırıldı. Bu rutin kontrollerde yakınması olmayan hasta 25,5 yılın sonunda baş ağrısı, dengesiz yürüyüş, bulantı ve kusma ile hastanemize getirildi. Çekilen MRG'inde sağ serebellopontin köşesinde iki komponentli izointens kitle görüldü. Kitlenin mediyal komportmanı kontrast tutulumu göstermemişken dura ile ilişkili lateral komportmanı ise homojen olarak kontrast tutulum göstermiştir. Hastaya sağ suboksipital kraniyektomi yapılarak lateral kompartmanın total rezeksiyonu ve radyoterapiye bağlı gelişen fibroz dokudan şüphe edilen mediyal kompartmandan biyopsi alındı. Kitlenin lateral kompartmanı histopatolojik olarak fibroblastik menenjiom (WHO grade I) ve mediyal kompartmanın biyopsi sonucu araknoidal fibroz olarak tanı konulmuştur. Hasta 30 aylık takibinde iyidir nüks saptanmamıştır. Anahtar kelimeler: Radyasyonla ilişkili menenjiom, çocuklarda medülloblastom, radyoterapinin geç etkisi J Nervous Sys Surgery 2016;6(1-2):61-66
doi:10.5222/sscd.2016.72325 fatcat:b6upnj477zaxhf3vsmfl3lsqdm

Good Features to Correlate for Visual Tracking

Erhan Gundogdu, A. Aydin Alatan
2018 IEEE Transactions on Image Processing  
In order to improve the accuracy of the tracking, various machine learning concepts have been borrowed, such Erhan Gundogdu is with the Department of Intelligent Data Analytics Research Program Department  ... 
doi:10.1109/tip.2018.2806280 pmid:29994635 fatcat:p4ap5ibd5ngirlk5qcdxstgv2e

The effect of swallowing rehabilitation on quality of life of the dysphagic patients with cortical ischemic stroke

Kadir Bahcecı, Ebru Umay, Ibrahim Gundogdu, Eda Gurcay, Erhan Ozturk, Sibel Alıcura
2017 Iranian journal of neurology  
Swallowing and swallowing-related quality of life studies following stroke were almost always performed by including both patients with brainstem and cortical involvement. It was aimed in this study to show the presence of dysphagia in patients with only cortical ischemic stroke and to investigate the interaction between dysphagia and quality of life as well as to evaluate the effect of a rehabilitation program in the acute phase. Methods: Seventy-two patients with cortical stroke (between 0
more » ... 30 days) and dysphagia were included. Swallowing function of patients was assessed by dysphagia screen questionnaire and fiberoptic endoscopic assessment. Also, functional impairment and swallowing quality of life were assessed. The swallowing rehabilitation program for 4 weeks was given to all patients. Results: All patients demonstrated disorders related to oral phase (n = 69, 95.8%), pharyngeal phase (n = 4, 5.6%) or both phases. The swallowing function, swallowing quality of life and functional impairment were improved at the end of therapy. Conclusion: Swallowing quality of life is severely affected in cortical hemispheric stroke patients and can be improved with an early rehabilitation program.
pmid:29736223 pmcid:PMC5937003 fatcat:eg72tlgofbhbrbirifngr4cmge

Evaluation of Feature Channels for Correlation-Filter-Based Visual Object Tracking in Infrared Spectrum

Erhan Gundogdu, Aykut Koc, Berkan Solmaz, Riad I. Hammoud, A. Aydin Alatan
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
Correlation filters for visual object tracking in visible imagery has been well-studied. Most of the correlation-filterbased methods use either raw image intensities or feature maps of gradient orientations or color channels. However, well-known features designed for visible spectrum may not be ideal for infrared object tracking, since infrared and visible spectra have dissimilar characteristics in general. We assess the performance of two state-of-the-art correlationfilter-based object
more » ... methods on Linköping Thermal InfraRed (LTIR) dataset of medium wave and longwave infrared videos, using deep convolutional neural networks (CNN) features as well as other traditional hand-crafted descriptors. The deep CNN features are trained on an infrared dataset consisting of 16K objects for a supervised classification task. The highest performance in terms of the overlap metric is achieved when these deep CNN features are utilized in a correlation-filter-based tracker.
doi:10.1109/cvprw.2016.43 dblp:conf/cvpr/GundogduKSHA16 fatcat:q5zzxuvhh5b2hdai2f3h75sq4y

Bone mass and vitamin D levels in Parkinson's disease: is there any difference between genders?

Erhan Arif Ozturk, Ibrahim Gundogdu, Burak Tonuk, Bilge Gonenli Kocer, Yasemin Tombak, Selcuk Comoglu, Aytul Cakci
2016 Journal of Physical Therapy Science  
Purpose] The aim of this study was to determine the bone mineral density, vitamin D level, and frequencies of osteopenia and osteoporosis in patients with Parkinson's disease and to compare male and female patients with the controls separately. [Subjects and Methods] One hundred fifteen Parkinson's disease patients (47 males, 68 females; age range: 55-85 years) and 117 age-and gender-matched controls (47 males, 70 females) were enrolled in the study. Bone mineral density measured by dual-energy
more » ... X-ray absorptiometry and serum D vitamin levels of each participant were recorded. [Results] The mean lumbar spine, femur neck, and total femur bone mineral density levels, T-scores, and vitamin D levels were found to be significantly lower in Parkinson's disease patients in both genders. Furthermore, osteoporosis rates were found be significantly higher only in female Parkinson's disease patients compared with female controls. [Conclusion] Data from the present study revealed that while osteoporosis was significantly higher only in female Parkinson's disease patients, all Parkinson's disease patients had lower bone mineral density scores and vitamin D levels compared with the controls regardless of gender, suggesting that clinicians should pay attention to the osteoporosis risk in Parkinson's disease and that adequate preventive measures should be taken in order to limit the future risk due to osteoporotic fractures.
doi:10.1589/jpts.28.2204 pmid:27630398 pmcid:PMC5011562 fatcat:ah6ucdrjxzgnhddqs4q3fnrrj4

Sparse representation of two- and three-dimensional images with fractional Fourier, Hartley, linear canonical, and Haar wavelet transforms

Aykut Koç, Burak Bartan, Erhan Gundogdu, Tolga Çukur, Haldun M. Ozaktas
2017 Expert systems with applications  
Sparse recovery aims to reconstruct signals that are sparse in a linear transform domain from a heavily underdetermined set of measurements. The success of sparse recovery relies critically on the knowledge of transform domains that give compressible representations of the signal of interest. Here we consider two-and three-dimensional images, and investigate various multi-dimensional transforms in terms of the compressibility of the resultant coefficients. Specifically, we compare the
more » ... Fourier (FRT) and linear canonical transforms (LCT), which are generalized versions of the Fourier transform (FT), as well as Hartley and simplified fractional Hartley transforms, which differ from corresponding Fourier transforms in that they produce real outputs for real inputs. We also examine a cascade approach to improve transform-domain sparsity, where the Haar wavelet transform is applied following an initial Hartley transform. To compare the various methods, images are recovered from a subset of coefficients in the respective transform domains. The number of coefficients that are retained in the subset are varied systematically to examine the level of signal sparsity in each transform domain. Recovery performance is assessed via the structural similarity index (SSIM) and mean squared error (MSE) in reference to original images. Our analyses show that FRT and LCT transform yield the most sparse representations among the tested transforms as dictated by the improved quality of the recovered images. Furthermore, the cascade approach improves transform-domain sparsity among techniques applied on small image patches.
doi:10.1016/j.eswa.2017.01.046 fatcat:gg6ibw2es5gilglgzvpb4vyixu
« Previous Showing results 1 — 15 out of 91 results