Search

12th International Workshop on Machine Learning in Medical Imaging, MLMI 2021, held in conjunction with 24th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2021 -- 27 September 2021 through 27 September 2021 -- -- 266089, Visual explanation methods have an important role in the prognosis of the patients where the annotated data is limited or unavailable. There have been several attempts to use gradient-based attribution methods to localize pathology from medical scans without using segmentation labels. This research direction has been impeded by the lack of robustness and reliability. These methods are highly sensitive to the network parameters. In this study, we introduce a robust visual explanation method to address this problem for medical applications. We provide an innovative visual explanation algorithm for general purpose and as an example application we demonstrate its effectiveness for quantifying lesions in the lungs caused by the Covid-19 with high accuracy and robustness without using dense segmentation labels. This approach overcomes the drawbacks of commonly used Grad-CAM and its extended versions. The premise behind our proposed strategy is that the information flow is minimized while ensuring the classifier prediction stays similar. Our findings indicate that the bottleneck condition provides a more stable severity estimation than the similar attribution methods. The source code will be publicly available upon publication. © 2021, Springer Nature Switzerland AG.

18th IEEE International Conference on Intelligent Transportation Systems, ITSC 2015; Palacio de Congresos de Canarias Avenida Principe de Asturias Gran Canaria; Spain; 15 September 2015 through 18 September 2015, In this paper, we present our work on vehicle classification with omnidirectional cameras. In particular, we investigate whether the combined use of shape-based and gradient-based classifiers outperforms the individual classifiers or not. For shape-based classification, we extract features from the silhouettes in the omnidirectional video frames, which are obtained after background subtraction. Classification is performed with kNN (k Nearest Neighbors) method, which has been commonly used in shape-based vehicle classification studies in the past. For gradient-based classification, we employ HOG (Histogram of Oriented Gradients) features. Instead of searching a whole video frame, we extract the features in the region located by the foreground silhouette. We use SVM (Support Vector Machines) as the classifier since HOG+SVM is a commonly used pair in visual object detection. The vehicle types that we worked on are motorcycle, car and van (minibus). In experiments, we first analyze the performances of shape-based and HOG-based classifiers separately. Then, we analyze the performance of the combined classifier where the two classifiers are fused at decision level. Results show that the combined classifier is superior to the individual classifiers. © 2015 IEEE., TUBITAK (project 113E107)

In this paper, a recently introduced nonlinear gradient-based observer [1] has been adopted for Takagi-Sugeno (TS) fuzzy systems. The designed observer is especially aimed to estimate the unmeasurable states of the TS fuzzy systems where the LMI solution is not feasible to find the observer gains. The estimation of gradient observer is evaluated based on the Levenberg-Marquardt direction where the local convergence property is guaranteed using Lyapunov function approach. The numerical simulations present accurate estimation results for TS fuzzy nonlinear systems including a comparison with the conventional Extended Kalman Filter (EKF) yielding acceptable results. © 2015 IEEE.

Bu çalışmada, mülteci kamplarında bulunan yaşam alanlarının/barınakların (çadır, konteynır vb.) uydu ve hava görüntülerinden otomatik olarak sezimi ve sayımına yönelik bir yöntem tanıtılmaktadır. Mülteci kamplarındaki barınakların çok yüksek çözünürlüklü (VHR) görüntülerde dahi çok küçük boyuta sahip, belli bir alanda çok sık ve bazen karmaşık şekilde yerleştirilmiş ve arkaplan ile ayrıştırılması zor nesnelere denk geliyor olması otomatik barınak/çadır sezim ve sayımını zorlaştırmaktadır. Bahsedilen problemin çözümüne yönelik olarak çok sınıflı çizge-kesme (graph-cut) bölütlemesi ve gölge bilgisinin kullanıldığı özgün bir yöntem çalışma kapsamında önerilmektedir. Buna göre, ilk adımda çizge-kesme bölütleme yöntemi ve morfoloji işlemleri uygulanarak çadır sezimi yapılmıştır. Bu adımda sezilememiş çadırların bulunabilmesi için ikinci bir adımda çadırların gölge bilgisinden faydalanılmıştır. En son adımda gradyan-tabanlı eşikleme yöntemi ile yanlış tespitlerin elenmesi amaçlanmıştır. Önerilen yöntem, özellikle karmaşık yerleşim düzenine sahip mülteci kampları (çadırkent) üzerinde test edilmiştir. Yöntemin başarımı kesinlik (precision) ve geri getirme (recall) kriterlerine göre belirlenmiştir. Elde edilen ortalama kesinlik ve geri getirme başarımları sırasıyla %91.9 ve %90.0 olup problemin zorluğu göz önünde bulundurulduğunda umut vadedici bir sonuçtur. This paper adresses the automated detection and enumeration of the dwellings/tents within the refugee camp areas using satellite and aerial images. The fact that dwellings inside the refugee camps correspond to very small, densely and sometimes arbitrarily positioned objects that are hard to distinguish from the background, even in Very-High Resolution (VHR) images, makes the automated tent/dwelling detection a difficul problem. In this paper, a method combining the use of the multi-class graph-cut segmentation and shadow information is proposed for dwelling/tent detection. Accordingly, as a first step, tents/dwellings are detected by applying graph-cut segmentation and morphology. Then, in order to improve the accuracy, the shadow information is exploited. At the last step, all detected tent segments are re-checked for correctness using a gradient-based thresholding method. The proposed method is tested on the images of the refugee camps that mostly have arbitrary and dense dwelling/tent positioning inside. The precision and recall values are determined to evaluate the performance. The obtained average precision (91.9N and average recall (90.0%) performances are promising considering the difficulty of the problem. Publisher's Version

Video editing and object recognition are two significant fields in computer vi- sion: the first has remarkably assisted digital production and post-production tasks of a digital video footage; the second is considered fundamental to image classification or image based search in large databases (e.g. the web). In this thesis, we address two problems, namely we present a novel formulation that tackles video editing tasks and we develop a mechanism that allows to generate more robust descriptors for objects in an image. Concerning the first problem, this thesis proposes two variational models to perform temporally coherent video editing. These models are applied to change an object’s (rigid or non-rigid) texture throughout a given video sequence. One model is based on propagating color information from a given frame (or be- tween two given frames) along the motion trajectories of the video; while the other is based on propagating gradient domain information. The models we present in this thesis require minimal user intervention and they automatically accommodate for illumination changes in the scene. Concerning the second problem, this thesis addresses the problem of affine invariance in object recognition. We introduce a way to generate geometric affine invariant quantities that are used in the construction of feature descrip- tors. We show that when these quantities are used they do indeed achieve a more robust recognition than the state of the art descriptors. i, La edición de vídeo y el reconocimiento de objetos son dos áreas fundamentales en el campo de la visión por computador: la primera es de gran utilidad en los procesos de producción y post-producción digital de vídeo; la segunda es esencial para la clasificación o búsqueda de imágenes en grandes bases de datos (por ejemplo, en la web). En esta tesis se acometen ambos problemas, en concreto, se presenta una nueva formulación que aborda las tareas de edición de vídeo y se desarrolla un mecanismo que permite generar descriptores más robustos para los objetos de la imagen. Con respecto al primer problema, en esta tesis se proponen dos modelos variacionales para llevar a cabo la edición de vídeo de forma coherente en el tiempo. Estos modelos se aplican para cambiar la textura de un objeto (rígido o no) a lo largo de una secuencia de vídeo dada. Uno de los modelos está basado en la propagación de la información de color desde un determinado cuadro de la secuencia de vídeo (o entre dos cuadros dados) a lo largo de las trayectorias de movimiento del vídeo. El otro modelo está basado en la propagación de la información en el dominio del gradiente. Ambos modelos requieren una intervención mínima por parte del usuario y se ajustan de manera automática a los cambios de iluminación de la escena. Con respecto al segundo problema, esta tesis aborda el problema de la invariancia afín en el reconocimiento de objetos. Se introduce un nuevo método para generar cantidades geométricas afines que se utilizan en la generación de descriptores de características. También se demuestra que el uso de dichas cantidades proporciona mayor robustez al reconocimiento que los descriptores existentes actualmente en el estado del arte.

Date of Conference: 20-22 April 2011 Conference name: IEEE 19th Signal Processing and Communications Applications Conference (SIU), 2011 The first step of targeted cancer drug development is to screen and determine drug candidates by in vitro measuring the effectiveness of the drugs. The tests developed for this purpose can be time consuming due to their procedures and cannot be conducted in every laboratory due to the required hardwares. On the other hand, an image-based screening test has a potential to be less time consuming since it can directly be carried out on the live cell images and to be more extensively used because of the availability of its required equipments and their relatively less expensive cost. With such an image-based test, it is possible to quantify the cell death by finding cellular regions and comparing it against the control group. In this work, we propose a new method that automatically locates the cellular regions by the unsupervised segmentation of live cell images. This method relies on approximately locating cellular regions and the background with gradient-based thresholding and morphological operators and then finding the final boundaries by modeling the gradient of these regions with Gaussians. Working on the images of different cell lines captured with different magnifications, our experiments show that the proposed method leads to promising results. © 2011 IEEE. Kanser için hedefe yönelik ilaç geliştirme çalışmalarında ilk aşama, farklı antikanser maddelerin hücre kültürü ortamında etkinliğinin ölçülerek taranması ve ilaç adaylarının belirlenmesidir. Bu amaçla geliştirilen testler, prosedürleri gereği zaman alıcı olmakta ve ihtiyaç duydukları donanımdan dolayı her laboratuarda yapılamamaktadır. Öte yandan, mikroskop altında görüntüye dayalı yapılacak bir tarama testi, canlı hücre görüntülerinde ek işleme gerek duymadan uygulanabilir olmasından dolayı daha az zaman alma ve gerektirdiği ekipmanın bulunabilirliği ile nispeten daha az maliyetinden dolayı daha yaygın kullanılma potansiyeline sahiptir. Bu tip bir testle ilgili görüntüdeki hücresel bölgelerin kontrol grubu ile karşılaştırılıp hücre ölüm miktarlarının saptanması mümkündür. Bu çalışmada, canlı hücre görüntülerini öğreticisiz bölütleyerek hücresel bölgeleri otomatik saptayan bir yöntem geliştirilmiştir. Bu yöntem, öncelikle gradyan eşiklemesi ve morfolojik operatörlerle arka alan ve hücresel bölgelerin yaklaşık olarak bulunmasına, ardından bu bölgelerdeki renk ve gradyan bilgisinin Gauss tabanlı modellenerek son sınırların belirlenmesine dayanır. Farklı dokulara ait kanserli hücre hatlarından farklı büyütme kullanılarak alınmış görüntüler üzerinde yaptığımız çalışmalar, bu yöntemin hücresel bölgelerin yüksek doğrulukla saptanmasında umut verici sonuçlar verdiğini göstermiştir.

Data domain description techniques aim at deriving concise descriptions of objects belonging to a category of interest. For instance, the support vector domain description (SVDD) learns a hypersphere enclosing the bulk of provided unlabeled data such that points lying outside of the ball are considered anomalous. However, relevant information such as expert and background knowledge remain unused in the unsupervised setting. In this paper, we rephrase data domain description as a semi-supervised learning task, that is, we propose a semi-supervised generalization of data domain description (SSSVDD) to process unlabeled and labeled examples. The corresponding optimization problem is non-convex. We translate it into an unconstraint, continuous problem that can be optimized accurately by gradient-based techniques. Furthermore, we devise an effective active learning strategy to query low-confidence observations. Our empirical evaluation on network intrusion detection and object recognition tasks shows that our SSSVDDs consistently outperform baseline methods in relevant learning settings. Data domain description techniques aim at deriving concise descriptions of objects belonging to a category of interest. For instance, the support vector domain description (SVDD) learns a hypersphere enclosing the bulk of provided unlabeled data such that points lying outside of the ball are considered anomalous. However, relevant information such as expert and background knowledge remain unused in the unsupervised setting. In this paper, we rephrase data domain description as a semi-supervised learning task, that is, we propose a semi-supervised generalization of data domain description (SSSVDD) to process unlabeled and labeled examples. The corresponding optimization problem is non-convex. We translate it into an unconstraint, continuous problem that can be optimized accurately by gradient-based techniques. Furthermore, we devise an effective active learning strategy to query low-confidence observations. Our empirical evaluation on network intrusion detection and object recognition tasks shows that our SSSVDDs consistently outperform baseline methods in relevant learning settings.

An exact discrete adjoint of an unstructured nite-volume solver for the RANS equations has been developed. The adjoint is exact in the sense of being based on the full linearization of all terms in the solver, including all turbulence model contributions. From this starting point various approximations to the adjoint are derived with the intention of simplifying the development and memory requirements of the method; considered are many approximations already seen in the literature. The eect of these approximations on the accuracy of the resulting design gradients, and the convergence and nal solution of optimizations is studied, as it applies to a two-dimensional high-lift conguration.