Browsing by Author "Aksoy, Selim"

Now showing 1 - 20 of 63

Open Access
Automated detection of objects using multiple hierarchical segmentations
(IEEE, 2007-07) Akçay H. Gökhan; Aksoy, Selim
We introduce an unsupervised method that combines both spectral and structural information for automatic object detection. First, a segmentation hierarchy is constructed by combining structural information extracted by morphological processing with spectral information summarized using principal components analysis. Then, segments that maximize a measure consisting of spectral homogeneity and neighborhood connectivity are selected as candidate structures for object detection. Given the observation that different structures appear more clearly in different principal components, we present an algorithm that is based on probabilistic Latent Semantic Analysis (PLSA) for grouping the candidate segments belonging to multiple segmentations and multiple principal components. The segments are modeled using their spectral content and the PLSA algorithm builds object models by learning the object-conditional probability distributions. Labeling of a segment is done by computing the similarity of its spectral distribution to the distribution of object models using Kullback-Leibler divergence. Experiments on two data sets show that our method is able to automatically detect, group, and label segments belonging to the same object classes. © 2007 IEEE.
Open Access
Bağlamsal çıkarımla nesne sezimi
(IEEE, 2009-04) Kalaycılar, Fırat; Aksoy, Selim
Bu bildiride, sezim başarımını arttırmada tek tek sezilmiş nesneler arasındaki bağlamsal ilişkilerden yararlanan bir nesne sezim sistemi tanıtılmaktadır. Bu çalışmadaki ilk katkı, iki boyutlu görüntü uzayında yapılan ölçümlerden olasılıksal çıkarım yaparak nesneler arası gerçek dünya ilişkilerinin (çevresinde, yakınında, üzerinde vb.) modellenmesidir. Diğer bir katkı ise, bireysel nesne etiketlerine ve nesne ikilileri arasındaki ilişkilere bağlı olan sahne olasılık fonksiyonunun enbüyütülerek, nesnelerin en son etiketlerinin atanmasıdır. En tutarlı sahne duzenleşimini bulmak için bu enbüyütme problemi, doğrusal eniyileme kullanılarak çözülmüştür. Ofis görüntüleri içeren iki farklı veri kümesinde yapılan deneylerde, gerçek dünya uzamsal ilişkileri bağlamsal bilgi olarak kullanıldığında genel sezim başarımının arttığı gözlemlenmiştir. In this paper, an object detection system that utilizes contextual relationships between individually detected objects to improve the overall detection performance is introduced. The first contribution in this work is the modelling of real world object relationships (beside, on, near etc.) that can be probabilistically inferred using measurements in the 2D image space. The other contribution is the assignment offinol lobe/s to the detected objects by maximizing a scene probability function that is defined jointly using both individual object labels and their pairwise spatial relationships. The most consistent scene configuration is obtained by solving the maximization problem using linear optimization. Experiments on two different office data sets showed that incorporation of the real world spatial relationships as can textual information improved the overall detection performance. ©2009 IEEE.
Open Access
Bilkent University at TRECVID 2005
(National Institute of Standards and Technology, 2005-11) Aksoy, Selim; Avcı, Akın; Balçık, Erman; Çavuş, Özge; Duygulu, Pınar; Karaman, Zeynep; Kavak, Pınar; Kaynak, Cihan; Küçükayvaz, Emre; Öcalan, Çağdaş; Yıldız, Pınar
We describe our second-time participation, that includes one high-level feature extraction run, and three manual and one interactive search runs, to the TRECVID video retrieval evaluation. All of these runs have used a system trained on the common development collection. Only visual and textual information were used where visual information consisted of color, texture and edgebased low-level features and textual information consisted of the speech transcript provided in the collection. With the experience gained with our second-time participation, we are in the process of building a system for automatic classification and indexing of video archives.
Open Access
Bilkent university at TRECVID 2006
(National Institute of Standards and Technology, 2006-11) Aksoy, Selim; Duygulu, Pınar; Akçay, Hüseyin Gökhan; Ataer, Esra; Baştan, Muhammet; Can, Tolga; Çavuş, Özge; Doǧgrusöz, Emel; Gökalp, Demir; Akaydın, Ateş; Akoǧlu, Leman; Angın, Pelin; Cinbiş, R. Gökberk; Gür, Tunay; Ünlü, Mehmet
We describe our third participation, that includes one high-level feature extraction run, and two manual and one interactive search runs, to the TRECVID video retrieval evaluation. All of these runs have used a system trained on the common development collection. Only visual and textual information were used where visual information consisted of color, texture and edge-based low-level features and textual information consisted of the speech transcript provided in the collection.
Open Access
Bilkent University at TRECVID 2007
(National Institute of Standards and Technology, 2007) Aksoy, Selim; Duygulu, Pınar; Aksoy, C.; Aydin, E.; Gunaydin, D.; Hadimli, K.; Koç L.; Olgun, Y.; Orhan, C.; Yakin G.
We describe our fourth participation, that includes two high-level feature extraction runs, and one manual search run, to the TRECVID video retrieval evaluation. All of these runs have used a system trained on the common development collection. Only visual information, consisting of color, texture and edge-based low-level features, was used.
Open Access
Building detection using directional spatial constraints
(IEEE, 2010) Akçay, H. Gökhan; Aksoy, Selim
We propose an algorithm for automatic detection of buildings with complex shapes and roof structures in very high spatial resolution remotely sensed images. First, an initial oversegmentation is obtained. Then, candidate building regions are found using shadow and sun azimuth angle information. Finally, the building regions are selected by clustering the candidate regions using minimum spanning trees. The experiments on Ikonos scenes show that the algorithm is able to detect buildings with complex appearances and shapes. © 2010 IEEE.
Open Access
Çizge kesit yöntemi ile hiperspektral görüntülerde anomali tabanlı hedef tespiti
(IEEE, 2015-05) Batı, E.; Erdinç, Acar; Çeşmeci, D.; Çalışkan, A.; Koz, A.; Aksoy, Selim; Ertürk, S.; Alatan, A. A.
Hiperspektral hedef tespiti için yürütülen çalışmalar genel olarak iki sınıfta degerlendirilebilir. İlk sınıf olan anomali tespit yöntemlerinde, hedefin görüntünün geri kalanından farklı oldugu bilgisi kullanılarak görüntü analiz edilmektedir. Diğer sınıfta ise daha önceden bilgisi edinilmiş hedefe ait spektral imza ile görüntüdeki herbir piksel arasındaki benzerlik bulunarak hedefin konumu tespit edimektedir. Her iki sınıf yöntemin de önemli bir dezavantajı hiperspektral görüntü piksellerini bagımsız olarak degerlendirip, aralarındaki komşuluk ilişkilerini gözardı etmesidir. Bu makalede anomali tespit ve imza tabanlı tespit yakla¸sımlarını, pikseller arası komşuluk ilişkilerini de göz önünde bulundurarak birleştiren çizge yaklaşımına dayalı yeni bir yöntem önerilmiştir. Hedeflerin hem imza bilgisine sahip olundugu hem de anomali sayılabilecek ölçülerde olduğu varsayılarak önerilen çizge yaklaşımında önplan için imza bilgisi kullanan özgün bir türev tabanlı uyumlu filtre önerilmiştir. Arkaplan için ise seyreklik bilgisi kullanarak Gauss karışım bileşeni kestirimi yapan yeni bir anomali tespit yöntemi geliştirilmiştir. Son olarak komşular arası benzerligi tanımlamak için ise spektral bir benzerlik ölçütü olan spektral açı eleştiricisi kullanılmıştır. Önerilen çizge tabanlı yöntemin önplan, arkaplan ve komşuluk ilişkilerini uygun şekilde birleştirdigi ve önceki yöntemlere göre hedefi gürültüden arınmış bir bütün şeklinde başarıyla tespit edebildigi gözlemlenmiştir. The studies on hyperspectral target detection until now, has been treated in two approaches. Anomaly detection can be considered as the first approach, which analyses the hyperspectral image with respect to the difference between target and the rest of the hyperspectral image. The second approach compares the previously obtained spectral signature of the target with the pixels of the hyperspectral image in order to localize the target. A distinctive disadvantage of the aforementioned approaches is to treat each pixel of the hyperspectral image individually, without considering the neighbourhood relations between the pixels. In this paper, we propose a target detection algorithm which combines the anomaly detection and signature based hyperspectral target detection approaches in a graph based framework by utilizing the neighbourhood relations between the pixels. Assuming that the target signature is available and the target sizes are in the range of anomaly sizes, a novel derivative based matched filter is first proposed to model the foreground. Second, a new anomaly detection method which models the background as a Gaussian mixture is developed. The developed model estimates the optimal number of components forming the Gaussian mixture by means of utilizing sparsity information. Finally, the similarity of the neighbouring hyperspectral pixels is measured with the spectral angle mapper. The overall proposed graph based method has successfully combined the foreground, background and neighbouring information and improved the detection performance by locating the target as a whole object free from noises. © 2015 IEEE.
Open Access
The COST292 experimental framework for TRECVID 2007
(National Institute of Standards and Technology, 2007) Zhang, Q.; Corvaglia, M.; Aksoy, Selim; Naci, U.; Adami, N.; Aginako, N.; Alatan, A.; Alexandre, L. A.; Almeida, P.; Avrithis, Y.; Benois-Pineau, J.; Chandramouli, K.; Damnjanovic, U.; Esen, E.; Goya, J.; Grzegorzek, M.; Hanjalic, A.; Izquierdo, E.; Jarina, R.; Kapsalas, P.; Kompatsiaris, I.; Kuba, M.; Leonardi, R.; Makris, L.; Mansencal, B.; Mezaris, V.; Moumtzidou, A.; Mylonas, P.; Nikolopoulos, S.; Piatrik, T.; Pinheiro, A. M. G.; Reljin, B.; Spyrou, E.; Tolias, G.; Vrochidis, S.; Yakın, G.; Zajic, G.
In this paper, we give an overview of the four tasks submitted to TRECVID 2007 by COST292. In shot boundary (SB) detection task, four SB detectors have been developed and the results are merged using two merging algorithms. The framework developed for the high-level feature extraction task comprises four systems. The first system transforms a set of low-level descriptors into the semantic space using Latent Semantic Analysis and utilises neural networks for feature detection. The second system uses a Bayesian classifier trained with a "bag of subregions". The third system uses a multi-modal classifier based on SVMs and several descriptors. The fourth system uses two image classifiers based on ant colony optimisation and particle swarm optimisation respectively. The system submitted to the search task is an interactive retrieval application combining retrieval functionalities in various modalities with a user interface supporting automatic and interactive search over all queries submitted. Finally, the rushes task submission is based on a video summarisation and browsing system comprising two different interest curve algorithms and three features.
Open Access
Deep convolutional networks for PET super-resolution
(SPIE; Philips Res; Merck & Co Inc; Guerbet Grp; GE Res, 2024) Özaltan, Cemhan Kaan; Türkölmez, Emir; Namer,I. Jacques; Çiçek, A. Ercüment; Aksoy, Selim; Colliot, Olivier
Positron emission tomography (PET) provides valuable functional information that is widely used in clinical domains such as oncology and neurology. However, the structural quality of PET images may not be sufficient to effectively evaluate small regions of interest. Image super-resolution techniques aim to recover a high-resolution image from an input low-resolution version. We study adaptations of deep convolutional neural network architectures for improving the spatial resolution of PET images. The proposed super-resolution model involves a deep architecture that uses convolutional blocks together with various residual connections for more effective and efficient training. We use the supervised setting where the downscaled versions of the original PET images are given as the low-resolution input to the deep networks and the original images are used as the high-resolution target data to be recovered. Experiments show that the proposed model performs better than a multi-scale convolutional architecture according to both quantitative performance metrics and visual qualitative evaluation.
Open Access
Deep feature representations for variable-sized regions of ınterest in breast histopathology
(IEEE, 2021) Mercan, Caner; Aygüneş, Bulut; Aksoy, Selim; Mercan, Ezgi; Shapiro, L. G.; Weaver, D. L.; Elmore, J. G.
Objective: Modeling variable-sized regions of interest (ROIs) in whole slide images using deep convolutional networks is a challenging task, as these networks typically require fixed-sized inputs that should contain sufficient structural and contextual information for classification. We propose a deep feature extraction framework that builds an ROI-level feature representation via weighted aggregation of the representations of variable numbers of fixed-sized patches sampled from nuclei-dense regions in breast histopathology images. Methods: First, the initial patch-level feature representations are extracted from both fully-connected layer activations and pixel-level convolutional layer activations of a deep network, and the weights are obtained from the class predictions of the same network trained on patch samples. Then, the final patch-level feature representations are computed by concatenation of weighted instances of the extracted feature activations. Finally, the ROI-level representation is obtained by fusion of the patch-level representations by average pooling. Results: Experiments using a well-characterized data set of 240 slides containing 437 ROIs marked by experienced pathologists with variable sizes and shapes result in an accuracy score of 72.65% in classifying ROIs into four diagnostic categories that cover the whole histologic spectrum. Conclusion: The results show that the proposed feature representations are superior to existing approaches and provide accuracies that are higher than the average accuracy of another set of pathologists. Significance: The proposed generic representation that can be extracted from any type of deep convolutional architecture combines the patch appearance information captured by the network activations and the diagnostic relevance predicted by the class-specific scoring of patches for effective modeling of variable-sized ROIs.
Open Access
Detection and classification of cancer in whole slide breast histopathology images using deep convolutional networks
(Elsevier, 2018) Geçer, Barış; Aksoy, Selim; Mercan, E.; Shapiro, L. G.; Weaver, D. L.; Elmore, J. G.
Generalizability of algorithms for binary cancer vs. no cancer classification is unknown for clinically more significant multi-class scenarios where intermediate categories have different risk factors and treatment strategies. We present a system that classifies whole slide images (WSI) of breast biopsies into five diagnostic categories. First, a saliency detector that uses a pipeline of four fully convolutional networks, trained with samples from records of pathologists’ screenings, performs multi-scale localization of diagnostically relevant regions of interest in WSI. Then, a convolutional network, trained from consensus-derived reference samples, classifies image patches as non-proliferative or proliferative changes, atypical ductal hyperplasia, ductal carcinoma in situ, and invasive carcinoma. Finally, the saliency and classification maps are fused for pixel-wise labeling and slide-level categorization. Experiments using 240 WSI showed that both saliency detector and classifier networks performed better than competing algorithms, and the five-class slide-level accuracy of 55% was not statistically different from the predictions of 45 pathologists. We also present example visualizations of the learned representations for breast cancer diagnosis.
Open Access
Detection of compound structures by region group selection from hierarchical segmentations
(IEEE, 2016-07) Akçay, H. Gökhan; Aksoy, Selim
Detection of compound structures that are comprised of different arrangements of simpler primitive objects has been a challenging problem as commonly used bag-of-words models are limited in capturing spatial information. We have developed a generic method that considers the primitive objects as random variables, builds a contextual model of their arrangements using a Markov random field, and detects new instances of compound structures through automatic selection of subsets of candidate regions from a hierarchical segmentation by maximizing the likelihood of their individual appearances and relative spatial arrangements. In this paper, we extend the model to handle different types of primitive objects that come from multiple hierarchical segmentations. Results are shown for the detection of different types of housing estates in a WorldView-2 image. © 2016 IEEE.
Open Access
Detection of compound structures using clustering of statistical and structural features
(IEEE, 2012) Akçay, H. Gökhan; Aksoy, Selim
We describe a new method for detecting compound structures in images by combining the statistical and structural characteristics of simple primitive objects. A graph is constructed by assigning the primitive objects to its vertices, and connecting potentially related objects using edges. Statistical information that is modeled using spectral, shape, and position data of individual objects as well as the structural information that is modeled in terms of spatial alignments of neighboring object groups are also encoded in this graph. Experiments using WorldView-2 data show that hierarchical clustering of the graph vertices can discover high-level compound structures that cannot be obtained using traditional techniques. © 2012 IEEE.
Open Access
Detection of compound structures using hierarchical clustering of statistical and structural features
(IEEE, 2011) Akçay, H. Gokhan; Aksoy, Selim
We describe a new procedure that combines statistical and structural characteristics of simple primitive objects to discover compound structures in images. The statistical information that is modeled using spectral, shape, and position data of individual objects, and structural information that is modeled in terms of spatial alignments of neighboring object groups are encoded in a graph structure that contains the primitive objects at its vertices, and the edges connect the potentially related objects. Experiments using WorldView-2 data show that hierarchical clustering of these vertices can find high-level compound structures that cannot be obtained using traditional techniques. © 2011 IEEE.
Open Access
Detection of compound structures using multiple hierarchical segmentations
(IEEE, 2014) Akçay, Hüseyin Gökhan; Aksoy, Selim
In this paper, we present a method for automatic compound structure detection in high-resolution images. Given a query compound structure, our aim is to detect coherent regions with similar spatial arrangement and characteristics in multiple hierarchical segmentations. A Markov random field is constructed by representing query regions as variables and connecting the vertices that are spatially close by edges. Then, a maximum entropy distribution is assumed over the query region process and selection of similar region processes among a set of region hierarchies is achieved by maximizing the query model. Experiments using WorldView-2 images show the efficiency of probabilistic modeling of compound structures. © 2014 IEEE.
Open Access
Detection of compound structures using multiple hierarchical segmentations
(IEEE, 2012) Akçay, H. Gökhan; Aksoy, Selim
In this paper, our aim is to discover compound structures comprised of regions obtained from hierarchical segmentations of multiple spectral bands. A region adjacency graph is constructed by representing regions as vertices and connecting these vertices that are spatially close by edges. Then, dissimilarities between neighboring vertices are computed using statistical and structural features, and are assigned as edge weights. Finally, the compound structures are detected by extracting the connected components of the graph whose edges with relatively large weights are removed. Experiments using WorldView-2 images show that grouping of these vertices according to different criteria can extract high-level compound structures that cannot be obtained using traditional techniques. © 2012 IEEE.
Open Access
Detection of heterogeneous structures using hierarchical segmentation
(IEEE, 2011) Akçay, H. Gokhan; Aksoy, Selim
We present an unsupervised hierarchical segmentation algorithm for detecting complex heterogeneous image structures that are comprised of simpler homogeneous primitive objects. The first step segments primitive objects with uniform spectral content. Then, the co-occurrence information between neighboring regions is modeled and clustered. We assume that dense clusters of this co-occurrence space can be considered significant. Finally, the neighboring regions within these clusters are merged to obtain the next level in the segmentation hierarchy. The experiments show that the algorithm that iteratively clusters and merges region groups is able to segment heterogeneous structures in a hierarchical manner. © 2011 IEEE.
Open Access
E-museum: web-based tour and information system for museums
(IEEE, 2006) Baştanlar, Y.; Altıngövde, İsmail Şenol; Aksay, A.; Alav, O.; Çavuş, Özge; Yardımcı, Y.; Ulusoy, Ozgur; Güdükbay, Uğur; Çetin, A. Enis; Akar, G. B.; Aksoy, Selim
A web-based system - consisting of data entrance, access and retrieval modules - is constructed for museums. Internet users that visit the e-museum, are able to view the written and visual information belonging to the artworks in the museum, are able to follow the virtual tour prepared for the different sections of the museum, are able to browse the artworks according to certain properties, are able to search the artworks having the similar visual content with the viewed artwork. © 2006 IEEE.
Open Access
Finding compound structures in images using image segmentation and graph-based knowledge discovery
(IEEE, 2009-07) Zamalieva, Daniya; Aksoy, Selim; Tilton J. C.
We present an unsupervised method for discovering compound image structures that are comprised of simpler primitive objects. An initial segmentation step produces image regions with homogeneous spectral content. Then, the segmentation is translated into a relational graph structure whose nodes correspond to the regions and the edges represent the relationships between these regions. We assume that the region objects that appear together frequently can be considered as strongly related. This relation is modeled using the transition frequencies between neighboring regions, and the significant relations are found as the modes of a probability distribution estimated using the features of these transitions. Experiments using an Ikonos image show that subgraphs found within the graph representing the whole image correspond to parts of different high-level compound structures. ©2009 IEEE.
Open Access
Fine-grained object recognition and zero-shot learning in multispectral imagery
(IEEE, 2018) Sümbül, Gencer; Aksoy, Selim; Cinbiş, R. G.
We present a method for fine-grained object recognition problem, that aims to recognize the type of an object among a large number of sub-categories, and zero-shot learning scenario on multispectral images. In order to establish a relation between seen classes and new unseen classes, a compatibility function between image features extracted from a convolutional neural network and auxiliary information of classes is learnt. Knowledge transfer for unseen classes is carried out by maximizing this function. Performance of the model (15.2%) evaluated with manually annotated attributes, a natural language model, and a scientific taxonomy as auxiliary information is promisingly better than the other methods for 16 test classes.