TY - BOOK
AB - Over the last years, kernel methods have established themselves as powerful tools for computer vision researchers as well as for practitioners. In this tutorial, we give an introduction to kernel methods in computer vision from a geometric perspective, introducing not only the ubiquitous support vector machines, but also less known techniques for regression, dimensionality reduction, outlier detection and clustering. Additionally, we give an outlook on very recent, non-classical techniques for the prediction of structure data, for the estimation of statistical dependency and for learning the kernel function itself. All methods are illustrated with examples of successful application from the recent computer vision research literature.
AU - Lampert, Christoph
ID - 3707
SN - 9781601982681
TI - Kernel methods in computer vision
VL - 4
ER -
TY - CONF
AB - Markov random field (MRF, CRF) models are popular in computer vision. However, in order to be computationally tractable they are limited to incorporate only local interactions and cannot model global properties, such as connectedness, which is a potentially useful high-level prior for object segmentation. In this work, we overcome this limitation by deriving a potential function that enforces the output labeling to be connected and that can naturally be used in the framework of recent MAP-MRF LP relaxations. Using techniques from polyhedral combinatorics, we show that a provably tight approximation to the MAP solution of the resulting MRF can still be found efficiently by solving a sequence of max-flow problems. The efficiency of the inference procedure also allows us to learn the parameters of a MRF with global connectivity potentials by means of a cutting plane algorithm. We experimentally evaluate our algorithm on both synthetic data and on the challenging segmentation task of the PASCAL VOC 2008 data set. We show that in both cases the addition of a connectedness prior significantly reduces the segmentation error.
AU - Nowozin, Sebastian
AU - Christoph Lampert
ID - 3708
TI - Global connectivity potentials for random field models
ER -
TY - CONF
AB - We study the task of detecting the occurrence of objects in large image collections or in videos, a problem that combines aspects of content based image retrieval and object localization. While most previous approaches are either limited to special kinds of queries, or do not scale to large image sets, we propose a new method, efficient subimage retrieval (ESR), which is at the same time very flexible and very efficient. Relying on a two-layered branch-and-bound setup, ESR performs object-based image retrieval in sets of 100,000 or more images within seconds. An extensive evaluation on several datasets shows that ESR is not only very fast, but it also achieves detection accuracies that are on par with or superior to previously published methods for object-based image retrieval.
AU - Christoph Lampert
ID - 3709
TI - Detecting objects in large image collections and videos by efficient subimage retrieval
ER -
TY - JOUR
AB - Most successful object recognition systems rely on binary classification, deciding only if an object is present or not, but not providing information on the actual object location. To estimate the object‘s location, one can take a sliding window approach, but this strongly increases the computational cost because the classifier or similarity function has to be evaluated over a large set of candidate subwindows. In this paper, we propose a simple yet powerful branch and bound scheme that allows efficient maximization of a large class of quality functions over all possible subimages. It converges to a globally optimal solution typically in linear or even sublinear time, in contrast to the quadratic scaling of exhaustive or sliding window search. We show how our method is applicable to different object detection and image retrieval scenarios. The achieved speedup allows the use of classifiers for localization that formerly were considered too slow for this task, such as SVMs with a spatial pyramid kernel or nearest-neighbor classifiers based on the chi^2 distance. We demonstrate state-of-the-art localization performance of the resulting systems on the UIUC Cars data set, the PASCAL VOC 2006 data set, and in the PASCAL VOC 2007 competition.
AU - Christoph Lampert
AU - Blaschko,Matthew B
AU - Hofmann,Thomas
ID - 3710
IS - 12
JF - IEEE Transactions on Pattern Analysis and Machine Intelligence
TI - Efficient subwindow search: A branch and bound framework for object localization
VL - 31
ER -
TY - CONF
AB - An important cue to high level scene understanding is to analyze the objects in the scene and their behavior and interactions. In this paper, we study the problem of classification of activities in videos, as this is an integral component of any scene understanding system, and present a novel approach for recognizing human action categories in videos by combining information from appearance and motion of human body parts. Our approach is based on tracking human body parts by using mixture particle filters and then clustering the particles using local non - parametric clustering, hence associating a local set of particles to each cluster mode. The trajectory of these cluster modes provides the ldquomotionrdquo information and the ldquoappearancerdquo information is provided by the statistical information about the relative motion of these local set of particles over a number of frames. Later we use a ldquoBag of Wordsrdquo model to build one histogram per video sequence from the set of these robust appearance and motion descriptors. These histograms provide us characteristic information which helps us to discriminate among various human actions which ultimately helps us in better understanding of the complete scene. We tested our approach on the standard KTH and Weizmann human action datasets and the results were comparable to the state of the art methods. Additionally our approach is able to distinguish between activities that involve the motion of complete body from those in which only certain body parts move. In other words, our method discriminates well between activities with ldquoglobal body motionrdquo like running, jogging etc. and ldquolocal motionrdquo like waving, boxing etc.
AU - Dhillon, Paramveer S
AU - Nowozin, Sebastian
AU - Christoph Lampert
ID - 3711
TI - Combining appearance and motion for human action classification in videos
ER -
TY - JOUR
AB - Parent-of-origin-specific (imprinted) gene expression is regulated in Arabidopsis thaliana endosperm by cytosine demethylation of the maternal genome mediated by the DNA glycosylase DEMETER, but the extent of the methylation changes is not known. Here, we show that virtually the entire endosperm genome is demethylated, coupled with extensive local non-CG hypermethylation of small interfering RNA–targeted sequences. Mutation of DEMETER partially restores endosperm CG methylation to levels found in other tissues, indicating that CG demethylation is specific to maternal sequences. Endosperm demethylation is accompanied by CHH hypermethylation of embryo transposable elements. Our findings demonstrate extensive reconfiguration of the endosperm methylation landscape that likely reinforces transposon silencing in the embryo.
AU - Hsieh, Tzung-Fu
AU - Ibarra, Christian A.
AU - Silva, Pedro
AU - Zemach, Assaf
AU - Eshed-Williams, Leor
AU - Fischer, Robert L.
AU - ZILBERMAN, Daniel
ID - 9453
IS - 5933
JF - Science
KW - Multidisciplinary
SN - 0036-8075
TI - Genome-wide demethylation of Arabidopsis endosperm
VL - 324
ER -
TY - CONF
AB - We present a review of recent work on the mathematical aspects of the BCS gap equation, covering our results of Ref. 9 as well our recent joint work with Hamza and Solovej and with Frank and Naboko, respectively. In addition, we mention some related new results.
AU - Hainzl, Christian
AU - Robert Seiringer
ID - 2331
TI - Spectral properties of the BCS gap equation of superfluidity
ER -
TY - CONF
AB - We present a rigorous proof of the appearance of quantized vortices in dilute trapped Bose gases with repulsive two-body interactions subject to rotation, which was obtained recently in joint work with Elliott Lieb.14 Starting from the many-body Schrödinger equation, we show that the ground state of such gases is, in a suitable limit, well described by the nonlinear Gross-Pitaevskii equation. In the case of axially symmetric traps, our results show that the appearance of quantized vortices causes spontaneous symmetry breaking in the ground state.
AU - Robert Seiringer
ID - 2332
TI - Vortices and Spontaneous Symmetry Breaking in Rotating Bose Gases
ER -
TY - JOUR
AB - A lower bound is derived on the free energy (per unit volume) of a homogeneous Bose gas at density Q and temperature T. In the dilute regime, i.e., when a3 1, where a denotes the scattering length of the pair-interaction potential, our bound differs to leading order from the expression for non-interacting particles by the term 4πa(2 2}-[ - c]2+). Here, c(T) denotes the critical density for Bose-Einstein condensation (for the non-interacting gas), and [ · ]+ = max{ ·, 0} denotes the positive part. Our bound is uniform in the temperature up to temperatures of the order of the critical temperature, i.e., T ~ 2/3 or smaller. One of the key ingredients in the proof is the use of coherent states to extend the method introduced in [17] for estimating correlations to temperatures below the critical one.
AU - Robert Seiringer
ID - 2374
IS - 3
JF - Communications in Mathematical Physics
TI - Free energy of a dilute Bose gas: Lower bound
VL - 279
ER -
TY - JOUR
AB - We derive upper and lower bounds on the critical temperature Tc and the energy gap Ξ (at zero temperature) for the BCS gap equation, describing spin- 1 2 fermions interacting via a local two-body interaction potential λV(x). At weak coupling λ 1 and under appropriate assumptions on V(x), our bounds show that Tc ∼A exp(-B/λ) and Ξ∼C exp(-B/λ) for some explicit coefficients A, B, and C depending on the interaction V(x) and the chemical potential μ. The ratio A/C turns out to be a universal constant, independent of both V(x) and μ. Our analysis is valid for any μ; for small μ, or low density, our formulas reduce to well-known expressions involving the scattering length of V(x).
AU - Hainzl, Christian
AU - Robert Seiringer
ID - 2376
IS - 18
JF - Physical Review B - Condensed Matter and Materials Physics
TI - Critical temperature and energy gap for the BCS equation
VL - 77
ER -