Adriana Romero, Nicolas Ballas, Samira Ebrahimi Kahou, Antoine Chassang, Carlo Gatta, & Yoshua Bengio. (2015). FitNets: Hints for Thin Deep Nets. In 3rd International Conference on Learning Representations ICLR2015.
Abstract: While depth tends to improve network performances, it also makes gradient-based training more difficult since deeper networks tend to be more non-linear. The recently proposed knowledge distillation approach is aimed at obtaining small and fast-to-execute models, and it has shown that a student network could imitate the soft output of a larger teacher network or ensemble of networks. In this paper, we extend this idea to allow the training of a student that is deeper and thinner than the teacher, using not only the outputs but also the intermediate representations learned by the teacher as hints to improve the training process and final performance of the student. Because the student intermediate hidden layer will generally be smaller than the teacher's intermediate hidden layer, additional parameters are introduced to map the student hidden layer to the prediction of the teacher hidden layer. This allows one to train deeper students that can generalize better or run faster, a trade-off that is controlled by the chosen student capacity. For example, on CIFAR-10, a deep student network with almost 10.4 times less parameters outperforms a larger, state-of-the-art teacher network.
Keywords: Computer Science ; Learning; Computer Science ;Neural and Evolutionary Computing
|
Marc Bolaños, Maite Garolera, & Petia Radeva. (2015). Object Discovery using CNN Features in Egocentric Videos. In Pattern Recognition and Image Analysis, Proceedings of 7th Iberian Conference , ibPRIA 2015 (Vol. 9117, pp. 67–74). LNCS.
Abstract: Lifelogging devices based on photo/video are spreading faster everyday. This growth can represent great benefits to develop methods for extraction of meaningful information about the user wearing the device and his/her environment. In this paper, we propose a semi-supervised strategy for easily discovering objects relevant to the person wearing a first-person camera. The egocentric video sequence acquired by the camera, uses both the appearance extracted by means of a deep convolutional neural network and an object refill methodology that allow to discover objects even in case of small amount of object appearance in the collection of images. We validate our method on a sequence of 1000 egocentric daily images and obtain results with an F-measure of 0.5, 0.17 better than the state of the art approach.
Keywords: Object discovery; Egocentric videos; Lifelogging; CNN
|
Estefania Talavera, Mariella Dimiccoli, Marc Bolaños, Maedeh Aghaei, & Petia Radeva. (2015). R-clustering for egocentric video segmentation. In Pattern Recognition and Image Analysis, Proceedings of 7th Iberian Conference , ibPRIA 2015 (Vol. 9117, pp. 327–336). LNCS. Springer International Publishing.
Abstract: In this paper, we present a new method for egocentric video temporal segmentation based on integrating a statistical mean change detector and agglomerative clustering(AC) within an energy-minimization framework. Given the tendency of most AC methods to oversegment video sequences when clustering their frames, we combine the clustering with a concept drift detection technique (ADWIN) that has rigorous guarantee of performances. ADWIN serves as a statistical upper bound for the clustering-based video segmentation. We integrate both techniques in an energy-minimization framework that serves to disambiguate the decision of both techniques and to complete the segmentation taking into account the temporal continuity of video frames descriptors. We present experiments over egocentric sets of more than 13.000 images acquired with different wearable cameras, showing that our method outperforms state-of-the-art clustering methods.
Keywords: Temporal video segmentation; Egocentric videos; Clustering
|
Marc Bolaños, R. Mestre, Estefania Talavera, Xavier Giro, & Petia Radeva. (2015). Visual Summary of Egocentric Photostreams by Representative Keyframes. In IEEE International Conference on Multimedia and Expo ICMEW2015 (pp. 1–6).
Abstract: Building a visual summary from an egocentric photostream captured by a lifelogging wearable camera is of high interest for different applications (e.g. memory reinforcement). In this paper, we propose a new summarization method based on keyframes selection that uses visual features extracted bymeans of a convolutional neural network. Our method applies an unsupervised clustering for dividing the photostreams into events, and finally extracts the most relevant keyframe for each event. We assess the results by applying a blind-taste test on a group of 20 people who assessed the quality of the
summaries.
Keywords: egocentric; lifelogging; summarization; keyframes
|
|
Tadashi Araki, Nobutaka Ikeda, Nilanjan Dey, Sayan Chakraborty, Luca Saba, Dinesh Kumar, et al. (2015). A comparative approach of four different image registration techniques for quantitative assessment of coronary artery calcium lesions using intravascular ultrasound. CMPB - Computer Methods and Programs in Biomedicine, 118(2), 158–172.
|
R.A.Bendezu, E.Barba, E.Burri, D.Cisternas, Carolina Malagelada, Santiago Segui, et al. (2015). Intestinal gas content and distribution in health and in patients with functional gut symptoms. NEUMOT - Neurogastroenterology & Motility, 27(9), 1249–1257.
Abstract: BACKGROUND:
The precise relation of intestinal gas to symptoms, particularly abdominal bloating and distension remains incompletely elucidated. Our aim was to define the normal values of intestinal gas volume and distribution and to identify abnormalities in relation to functional-type symptoms.
METHODS:
Abdominal computed tomography scans were evaluated in healthy subjects (n = 37) and in patients in three conditions: basal (when they were feeling well; n = 88), during an episode of abdominal distension (n = 82) and after a challenge diet (n = 24). Intestinal gas content and distribution were measured by an original analysis program. Identification of patients outside the normal range was performed by machine learning techniques (one-class classifier). Results are expressed as median (IQR) or mean ± SE, as appropriate.
KEY RESULTS:
In healthy subjects the gut contained 95 (71, 141) mL gas distributed along the entire lumen. No differences were detected between patients studied under asymptomatic basal conditions and healthy subjects. However, either during a spontaneous bloating episode or once challenged with a flatulogenic diet, luminal gas was found to be increased and/or abnormally distributed in about one-fourth of the patients. These patients detected outside the normal range by the classifier exhibited a significantly greater number of abnormal features than those within the normal range (3.7 ± 0.4 vs 0.4 ± 0.1; p < 0.001).
CONCLUSIONS & INFERENCES:
The analysis of a large cohort of subjects using original techniques provides unique and heretofore unavailable information on the volume and distribution of intestinal gas in normal conditions and in relation to functional gastrointestinal symptoms.
|
G.Blasco, Simone Balocco, J.Puig, J.Sanchez-Gonzalez, W.Ricart, J.Daunis-I-Estadella, et al. (2015). Carotid pulse wave velocity by magnetic resonance imaging is increased in middle-aged subjects with the metabolic syndrome. ICJI - International Journal of Cardiovascular Imaging, 31(3), 603–612.
Abstract: Arterial pulse wave velocity (PWV), an independent predictor of cardiovascular disease, physiologically increases with age; however, growing evidence suggests metabolic syndrome (MetS) accelerates this increase. Magnetic resonance imaging (MRI) enables reliable noninvasive assessment of arterial stiffness by measuring arterial PWV in specific vascular segments. We investigated the association between the presence of MetS and its components with carotid PWV (cPWV) in asymptomatic subjects without diabetes. We assessed cPWV by MRI in 61 individuals (mean age, 55.3 ± 14.1 years; median age, 55 years): 30 with MetS and 31 controls with similar age, sex, body mass index, and LDL-cholesterol levels. The study population was dichotomized by the median age. To remove the physiological association between PWV and age, unpaired t tests and multiple regression analyses were performed using the residuals of the regression between PWV and age. cPWV was higher in middle-aged subjects with MetS than in those without (p = 0.001), but no differences were found in elder subjects (p = 0.313). cPWV was associated with diastolic blood pressure (r = 0.276, p = 0.033) and waist circumference (r = 0.268, p = 0.038). The presence of MetS was associated with increased cPWV regardless of age, sex, blood pressure, and waist (p = 0.007). The MetS components contributing independently to an increased cPWV were hypertension (p = 0.018) and hypertriglyceridemia (p = 0.002). The presence of MetS is associated with an increased cPWV in middle-aged subjects. In particular, hypertension and hypertriglyceridemia may contribute to early progression of carotid stiffness.
Keywords: Metabolic syndrome; Arterial stiffness; Pulse wave velocity; Carotid artery; Magnetic resonance
|
Lluis Garrido, M.Guerrieri, & Laura Igual. (2015). Image Segmentation with Cage Active Contours. TIP - IEEE Transactions on Image Processing, 24(12), 5557–5566.
Abstract: In this paper, we present a framework for image segmentation based on parametrized active contours. The evolving contour is parametrized according to a reduced set of control points that form a closed polygon and have a clear visual interpretation. The parametrization, called mean value coordinates, stems from the techniques used in computer graphics to animate virtual models. Our framework allows to easily formulate region-based energies to segment an image. In particular, we present three different local region-based energy terms: 1) the mean model; 2) the Gaussian model; 3) and the histogram model. We show the behavior of our method on synthetic and real images and compare the performance with state-of-the-art level set methods.
Keywords: Level sets; Mean value coordinates; Parametrized active contours; level sets; mean value coordinates
|
|
Marta Nuñez-Garcia, Sonja Simpraga, M.Angeles Jurado, Maite Garolera, Roser Pueyo, & Laura Igual. (2015). FADR: Functional-Anatomical Discriminative Regions for rest fMRI Characterization. In Machine Learning in Medical Imaging, Proceedings of 6th International Workshop, MLMI 2015, Held in Conjunction with MICCAI 2015 (pp. 61–68).
|
|
Chen Zhang, Maria del Mar Vila Muñoz, Petia Radeva, Roberto Elosua, Maria Grau, Angels Betriu, et al. (2015). Carotid Artery Segmentation in Ultrasound Images. In Computing and Visualization for Intravascular Imaging and Computer Assisted Stenting (CVII-STENT2015), Joint MICCAI Workshops.
|
Maedeh Aghaei, Mariella Dimiccoli, & Petia Radeva. (2015). Towards social interaction detection in egocentric photo-streams. In Proceedings of SPIE, 8th International Conference on Machine Vision , ICMV 2015 (Vol. 9875).
Abstract: Detecting social interaction in videos relying solely on visual cues is a valuable task that is receiving increasing attention in recent years. In this work, we address this problem in the challenging domain of egocentric photo-streams captured by a low temporal resolution wearable camera (2fpm). The major difficulties to be handled in this context are the sparsity of observations as well as unpredictability of camera motion and attention orientation due to the fact that the camera is worn as part of clothing. Our method consists of four steps: multi-faces localization and tracking, 3D localization, pose estimation and analysis of f-formations. By estimating pair-to-pair interaction probabilities over the sequence, our method states the presence or absence of interaction with the camera wearer and specifies which people are more involved in the interaction. We tested our method over a dataset of 18.000 images and we show its reliability on our considered purpose. © (2015) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
|
Adriana Romero. (2015). Assisting the training of deep neural networks with applications to computer vision (Carlo Gatta, & Petia Radeva, Eds.). Ph.D. thesis, Ediciones Graficas Rey, .
Abstract: Deep learning has recently been enjoying an increasing popularity due to its success in solving challenging tasks. In particular, deep learning has proven to be effective in a large variety of computer vision tasks, such as image classification, object recognition and image parsing. Contrary to previous research, which required engineered feature representations, designed by experts, in order to succeed, deep learning attempts to learn representation hierarchies automatically from data. More recently, the trend has been to go deeper with representation hierarchies.
Learning (very) deep representation hierarchies is a challenging task, which
involves the optimization of highly non-convex functions. Therefore, the search
for algorithms to ease the learning of (very) deep representation hierarchies from data is extensive and ongoing.
In this thesis, we tackle the challenging problem of easing the learning of (very) deep representation hierarchies. We present a hyper-parameter free, off-the-shelf, simple and fast unsupervised algorithm to discover hidden structure from the input data by enforcing a very strong form of sparsity. We study the applicability and potential of the algorithm to learn representations of varying depth in a handful of applications and domains, highlighting the ability of the algorithm to provide discriminative feature representations that are able to achieve top performance.
Yet, while emphasizing the great value of unsupervised learning methods when
labeled data is scarce, the recent industrial success of deep learning has revolved around supervised learning. Supervised learning is currently the focus of many recent research advances, which have shown to excel at many computer vision tasks. Top performing systems often involve very large and deep models, which are not well suited for applications with time or memory limitations. More in line with the current trends, we engage in making top performing models more efficient, by designing very deep and thin models. Since training such very deep models still appears to be a challenging task, we introduce a novel algorithm that guides the training of very thin and deep models by hinting their intermediate representations.
Very deep and thin models trained by the proposed algorithm end up extracting feature representations that are comparable or even better performing
than the ones extracted by large state-of-the-art models, while compellingly
reducing the time and memory consumption of the model.
|
Julie Digne, Mariella Dimiccoli, Neus Sabater, & Philippe Salembier. (2015). Neighborhood Filters and the Recovery of 3D Information. In Handbook of Mathematical Methods in Imaging (pp. 1645–1673). Springer New York.
Abstract: Following their success in image processing (see Chapter Local Smoothing Neighborhood Filters), neighborhood filters have been extended to 3D surface processing. This adaptation is not straightforward. It has led to several variants for surfaces depending on whether the surface is defined as a mesh, or as a raw data point set. The image gray level in the bilateral similarity measure is replaced by a geometric information such as the normal or the curvature. The first section of this chapter reviews the variants of 3D mesh bilateral filters and compares them to the simplest possible isotropic filter, the mean curvature motion.In a second part, this chapter reviews applications of the bilateral filter to a data composed of a sparse depth map (or of depth cues) and of the image on which they have been computed. Such sparse depth cues can be obtained by stereovision or by psychophysical techniques. The underlying assumption to these applications is that pixels with similar intensity around a region are likely to have similar depths. Therefore, when diffusing depth information with a bilateral filter based on locality and color similarity, the discontinuities in depth are assured to be consistent with the color discontinuities, which is generally a desirable property. In the reviewed applications, this ends up with the reconstruction of a dense perceptual depth map from the joint data of an image and of depth cues.
|
Maedeh Aghaei, Mariella Dimiccoli, & Petia Radeva. (2015). Multi-Face Tracking by Extended Bag-of-Tracklets in Egocentric Videos.
Abstract: Egocentric images offer a hands-free way to record daily experiences and special events, where social interactions are of special interest. A natural question that arises is how to extract and track the appearance of multiple persons in a social event captured by a wearable camera. In this paper, we propose a novel method to find correspondences of multiple-faces in low temporal resolution egocentric sequences acquired through a wearable camera. This kind of sequences imposes additional challenges to the multitracking problem with respect to conventional videos. Due to the free motion of the camera and to its low temporal resolution (2 fpm), abrupt changes in the field of view, in illumination conditions and in the target location are very frequent. To overcome such a difficulty, we propose to generate, for each detected face, a set of correspondences along the whole sequence that we call tracklet and to take advantage of their redundancy to deal with both false positive face detections and unreliable tracklets. Similar tracklets are grouped into the so called extended bag-of-tracklets (eBoT), which are aimed to correspond to specific persons. Finally, a prototype tracklet is extracted for each eBoT. We validated our method over a dataset of 18.000 images from 38 egocentric sequences with 52 trackable persons and compared to the state-of-the-art methods, demonstrating its effectiveness and robustness.
|
Jordina Torrents-Barrena, Aida Valls, Petia Radeva, Meritxell Arenas, & Domenec Puig. (2015). Automatic Recognition of Molecular Subtypes of Breast Cancer in X-Ray images using Segmentation-based Fractal Texture Analysis. In Artificial Intelligence Research and Development (Vol. 277, pp. 247–256). Frontiers in Artificial Intelligence and Applications. IOS Press.
Abstract: Breast cancer disease has recently been classified into four subtypes regarding the molecular properties of the affected tumor region. For each patient, an accurate diagnosis of the specific type is vital to decide the most appropriate therapy in order to enhance life prospects. Nowadays, advanced therapeutic diagnosis research is focused on gene selection methods, which are not robust enough. Hence, we hypothesize that computer vision algorithms can offer benefits to address the problem of discriminating among them through X-Ray images. In this paper, we propose a novel approach driven by texture feature descriptors and machine learning techniques. First, we segment the tumour part through an active contour technique and then, we perform a complete fractal analysis to collect qualitative information of the region of interest in the feature extraction stage. Finally, several supervised and unsupervised classifiers are used to perform multiclass classification of the aforementioned data. The experimental results presented in this paper support that it is possible to establish a relation between each tumor subtype and the extracted features of the patterns revealed on mammograms.
|
