arXiv subjects

Christoph Mayer

Publications and source records attributed to Christoph Mayer.

18 recordsLinked to original sources

arXiv2025

LpBound: Pessimistic Cardinality Estimation using $\ell_p$-Norms of Degree Sequences

Cardinality estimation is the problem of estimating the size of the output of a query, without actually evaluating the query. The cardinality estimator is a critical piece of a query optimizer, and is often the main culprit when the optimizer chooses a poor plan. This paper introduces LpBound, a pessimistic cardinality estimator for multijoin queries (acyclic or cyclic) with selection predicates and group-by clauses. LpBound computes a guaranteed upper bound on the size of the query output using simple statistics on the input relations, consisting of $\ell_p$-norms of degree sequences. The bound is the optimal solution of a linear program whose constraints encode data statistics and Shannon inequalities. We introduce two optimizations that exploit the structure of the query in order to speed up the estimation time and make LpBound practical. We experimentally evaluate LpBound against a range of traditional, pessimistic, and machine learning-based estimators on the JOB, STATS, and subgraph matching benchmarks. Our main finding is that LpBound can be orders of magnitude more accurate than traditional estimators used in mainstream open-source and commercial database systems. Yet it has comparable low estimation time and space requirements. When injected the estimates of LpBound, Postgres derives query plans at least as good as those derived using the true cardinalities.

cs.DB↗

arXiv2022

Beyond SOT: Tracking Multiple Generic Objects at Once

Generic Object Tracking (GOT) is the problem of tracking target objects, specified by bounding boxes in the first frame of a video. While the task has received much attention in the last decades, researchers have almost exclusively focused on the single object setting. Multi-object GOT benefits from a wider applicability, rendering it more attractive in real-world applications. We attribute the lack of research interest into this problem to the absence of suitable benchmarks. In this work, we introduce a new large-scale GOT benchmark, LaGOT, containing multiple annotated target objects per sequence. Our benchmark allows users to tackle key remaining challenges in GOT, aiming to increase robustness and reduce computation through joint tracking of multiple objects simultaneously. In addition, we propose a transformer-based GOT tracker baseline capable of joint processing of multiple objects through shared computation. Our approach achieves a 4x faster run-time in case of 10 concurrent objects compared to tracking each object independently and outperforms existing single object trackers on our new benchmark. In addition, our approach achieves highly competitive results on single-object GOT datasets, setting a new state of the art on TrackingNet with a success rate AUC of 84.4%. Our benchmark, code, and trained models will be made publicly available.

cs.CV↗

arXiv2022

AVisT: A Benchmark for Visual Object Tracking in Adverse Visibility

One of the key factors behind the recent success in visual tracking is the availability of dedicated benchmarks. While being greatly benefiting to the tracking research, existing benchmarks do not pose the same difficulty as before with recent trackers achieving higher performance mainly due to (i) the introduction of more sophisticated transformers-based methods and (ii) the lack of diverse scenarios with adverse visibility such as, severe weather conditions, camouflage and imaging effects. We introduce AVisT, a dedicated benchmark for visual tracking in diverse scenarios with adverse visibility. AVisT comprises 120 challenging sequences with 80k annotated frames, spanning 18 diverse scenarios broadly grouped into five attributes with 42 object categories. The key contribution of AVisT is diverse and challenging scenarios covering severe weather conditions such as, dense fog, heavy rain and sandstorm; obstruction effects including, fire, sun glare and splashing water; adverse imaging effects such as, low-light; target effects including, small targets and distractor objects along with camouflage. We further benchmark 17 popular and recent trackers on AVisT with detailed analysis of their tracking performance across attributes, demonstrating a big room for improvement in performance. We believe that AVisT can greatly benefit the tracking community by complementing the existing benchmarks, in developing new creative tracking solutions in order to continue pushing the boundaries of the state-of-the-art. Our dataset along with the complete tracking performance evaluation is available at: https://github.com/visionml/pytracking

cs.CV↗

arXiv2022

Robust Visual Tracking by Segmentation

Estimating the target extent poses a fundamental challenge in visual object tracking. Typically, trackers are box-centric and fully rely on a bounding box to define the target in the scene. In practice, objects often have complex shapes and are not aligned with the image axis. In these cases, bounding boxes do not provide an accurate description of the target and often contain a majority of background pixels. We propose a segmentation-centric tracking pipeline that not only produces a highly accurate segmentation mask, but also internally works with segmentation masks instead of bounding boxes. Thus, our tracker is able to better learn a target representation that clearly differentiates the target in the scene from background content. In order to achieve the necessary robustness for the challenging tracking scenario, we propose a separate instance localization component that is used to condition the segmentation decoder when producing the output mask. We infer a bounding box from the segmentation mask, validate our tracker on challenging tracking datasets and achieve the new state of the art on LaSOT with a success AUC score of 69.7%. Since most tracking datasets do not contain mask annotations, we cannot use them to evaluate predicted segmentation masks. Instead, we validate our segmentation quality on two popular video object segmentation datasets.

cs.CV↗

arXiv2022

Transforming Model Prediction for Tracking

Optimization based tracking methods have been widely successful by integrating a target model prediction module, providing effective global reasoning by minimizing an objective function. While this inductive bias integrates valuable domain knowledge, it limits the expressivity of the tracking network. In this work, we therefore propose a tracker architecture employing a Transformer-based model prediction module. Transformers capture global relations with little inductive bias, allowing it to learn the prediction of more powerful target models. We further extend the model predictor to estimate a second set of weights that are applied for accurate bounding box regression. The resulting tracker relies on training and on test frame information in order to predict all weights transductively. We train the proposed tracker end-to-end and validate its performance by conducting comprehensive experiments on multiple tracking datasets. Our tracker sets a new state of the art on three benchmarks, achieving an AUC of 68.5% on the challenging LaSOT dataset.

cs.CV↗

arXiv2021

Learning Target Candidate Association to Keep Track of What Not to Track

The presence of objects that are confusingly similar to the tracked target, poses a fundamental challenge in appearance-based visual tracking. Such distractor objects are easily misclassified as the target itself, leading to eventual tracking failure. While most methods strive to suppress distractors through more powerful appearance models, we take an alternative approach. We propose to keep track of distractor objects in order to continue tracking the target. To this end, we introduce a learned association network, allowing us to propagate the identities of all target candidates from frame-to-frame. To tackle the problem of lacking ground-truth correspondences between distractor objects in visual tracking, we propose a training strategy that combines partial annotations with self-supervision. We conduct comprehensive experimental validation and analysis of our approach on several challenging datasets. Our tracker sets a new state-of-the-art on six benchmarks, achieving an AUC score of 67.1% on LaSOT and a +5.8% absolute gain on the OxUvA long-term dataset.

cs.CV↗

arXiv2020

Group Sparsity: The Hinge Between Filter Pruning and Decomposition for Network Compression

In this paper, we analyze two popular network compression techniques, i.e. filter pruning and low-rank decomposition, in a unified sense. By simply changing the way the sparsity regularization is enforced, filter pruning and low-rank decomposition can be derived accordingly. This provides another flexible choice for network compression because the techniques complement each other. For example, in popular network architectures with shortcut connections (e.g. ResNet), filter pruning cannot deal with the last convolutional layer in a ResBlock while the low-rank decomposition methods can. In addition, we propose to compress the whole network jointly instead of in a layer-wise manner. Our approach proves its potential as it compares favorably to the state-of-the-art on several benchmarks.

cs.CV↗

arXiv2019

Efficient Video Semantic Segmentation with Labels Propagation and Refinement

This paper tackles the problem of real-time semantic segmentation of high definition videos using a hybrid GPU / CPU approach. We propose an Efficient Video Segmentation(EVS) pipeline that combines: (i) On the CPU, a very fast optical flow method, that is used to exploit the temporal aspect of the video and propagate semantic information from one frame to the next. It runs in parallel with the GPU. (ii) On the GPU, two Convolutional Neural Networks: A main segmentation network that is used to predict dense semantic labels from scratch, and a Refiner that is designed to improve predictions from previous frames with the help of a fast Inconsistencies Attention Module (IAM). The latter can identify regions that cannot be propagated accurately. We suggest several operating points depending on the desired frame rate and accuracy. Our pipeline achieves accuracy levels competitive to the existing real-time methods for semantic image segmentation(mIoU above 60%), while achieving much higher frame rates. On the popular Cityscapes dataset with high resolution frames (2048 x 1024), the proposed operating points range from 80 to 1000 Hz on a single GPU and CPU.

cs.CV↗

arXiv2019

Adversarial Feature Distribution Alignment for Semi-Supervised Learning

Training deep neural networks with only a few labeled samples can lead to overfitting. This is problematic in semi-supervised learning where only a few labeled samples are available. In this paper, we show that a consequence of overfitting in SSL is feature distribution misalignment between labeled and unlabeled samples. Hence, we propose a new feature distribution alignment method. Our method is particularly effective when using only a small amount of labeled samples. We test our method on CIFAR10 and SVHN. On SVHN we achieve a test error of 3.88% (250 labeled samples) and 3.39% (1000 labeled samples) which is close to the fully supervised model 2.89% (73k labeled samples). In comparison, the current SOTA achieves only 4.29% and 3.74%. Finally, we provide a theoretical insight why feature distribution alignment occurs and show that our method reduces it.

cs.CV↗

arXiv2019

Adversarial Sampling for Active Learning

This paper proposes asal, a new GAN based active learning method that generates high entropy samples. Instead of directly annotating the synthetic samples, ASAL searches similar samples from the pool and includes them for training. Hence, the quality of new samples is high and annotations are reliable. To the best of our knowledge, ASAL is the first GAN based AL method applicable to multi-class problems that outperforms random sample selection. Another benefit of ASAL is its small run-time complexity (sub-linear) compared to traditional uncertainty sampling (linear). We present a comprehensive set of experiments on multiple traditional data sets and show that ASAL outperforms similar methods and clearly exceeds the established baseline (random sampling). In the discussion section we analyze in which situations ASAL performs best and why it is sometimes hard to outperform random sample selection.

cs.LG↗

arXiv2019

Towards Closing the Gap in Weakly Supervised Semantic Segmentation with DCNNs: Combining Local and Global Models

Generating training sets for deep convolutional neural networks (DCNNs) is a bottleneck for modern real-world applications. This is a demanding task for applications where annotating training data is costly, such as in semantic segmentation. In the literature, there is still a gap between the performance achieved by a network trained on full and on weak annotations. In this paper, we establish a strategy to measure this gap and to identify the ingredients necessary to reduce it. On scribbles, we establish new state-of-the-art results: we obtain a mIoU of 75.6% without, and 75.7% with CRF post-processing. We reduce the gap by 64.2% whereas the current state-of-the-art reduces it only by 57.5%. Thanks to a systematic study of the different ingredients involved in the weakly supervised scenario and an original experimental strategy, we unravel a counter-intuitive mechanism that is simple and amenable to generalisations to other weakly-supervised scenarios: averaging poor local predicted annotations with the baseline ones and reuse them for training a DCNN yields new state-of-the-art results.

cs.CV↗

arXiv2009

Lie Algebroid Yang Mills with Matter Fields

Lie algebroid Yang-Mills theories are a generalization of Yang-Mills gauge theories, replacing the structural Lie algebra by a Lie algebroid E. In this note we relax the conditions on the fiber metric of E for gauge invariance of the action functional. Coupling to scalar fields requires possibly nonlinear representations of Lie algebroids. In all cases, gauge invariance is seen to lead to a condition of covariant constancy on the respective fiber metric in question with respect to an appropriate Lie algebroid connection. The presentation is kept in part explicit so as to be accessible also to a less mathematically oriented audience.

hep-th↗

arXiv2005

Special geometry of Euclidean supersymmetry II: hypermultiplets and the c-map

We construct two new versions of the c-map which allow us to obtain the target manifolds of hypermultiplets in Euclidean theories with rigid N =2 supersymmetry. While the Minkowskian para-c-map is obtained by dimensional reduction of the Minkowskian vector multiplet lagrangian over time, the Euclidean para-c-map corresponds to the dimensional reduction of the Euclidean vector multiplet lagrangian. In both cases the resulting hypermultiplet target spaces are para-hyper-Kahler manifolds. We review and prove the relevant results of para-complex and para-hypercomplex geometry. In particular, we give a second, purely geometrical construction of both c-maps, by proving that the cotangent bundle N=T^*M of any affine special (para-)Kahler manifold M is para-hyper-Kahler.

hep-th↗

arXiv2004

Domain Walls, Hitchin's Flow Equations and G_2-Manifolds

We construct BPS domain wall solutions of the effective action of type-IIA string theory compactified on a half-flat six-manifold. The flow equations for the vector and hypermultiplet scalars are shown to be equivalent to Hitchin's flow equations, implying that our domain walls can be lifted to solutions of ten-dimensional type-IIA supergravity. They take the form R^{1,2} x Y_7, where Y_7 is a G_2-holonomy manifold with boundaries.

hep-th↗

arXiv2003

Space-Time Singularities and the Kahler Cone

We review recent results on the interplay between the five-dimensional space-time and the internal manifold in Calabi-Yau compactifications of M-theory. Black string, black hole and domain wall solutions as well as Kasner type cosmologies cannot develop a naked singularity as long as the moduli take values inside the Kahler cone.

hep-th↗

arXiv2003

The Kahler Cone as Cosmic Censor

M-theory effects prevent five-dimensional domain-wall and black-hole solutions from developing curvature singularities. While so far this analysis was performed for particular models, we now present a model-independent proof that these solutions do not have naked singularities as long as the Kahler moduli take values inside the extended Kahler cone. As a by-product we obtain information on the regularity of the Kahler-cone metric at boundaries of the Kahler cone and derive relations between the geometry of moduli space and space-time.

hep-th↗

arXiv2003

Special Geometry of Euclidean Supersymmetry I: Vector Multiplets

We construct the general action for Abelian vector multiplets in rigid 4-dimensional Euclidean (instead of Minkowskian) N=2 supersymmetry, i.e., over space-times with a positive definite instead of a Lorentzian metric. The target manifolds for the scalar fields turn out to be para-complex manifolds endowed with a particular kind of special geometry, which we call affine special para-Kahler geometry. We give a precise definition and develop the mathematical theory of such manifolds. The relation to the affine special Kahler manifolds appearing in Minkowskian N=2 supersymmetry is discussed. Starting from the general 5-dimensional vector multiplet action we consider dimensional reduction over time and space in parallel, providing a dictionary between the resulting Euclidean and Minkowskian theories. Then we reanalyze supersymmetry in four dimensions and find that any (para-)holomorphic prepotential defines a supersymmetric Lagrangian, provided that we add a specific four-fermion term, which cannot be obtained by dimensional reduction. We show that the Euclidean action and supersymmetry transformations, when written in terms of para-holomorphic coordinates, take exactly the same form as their Minkowskian counterparts. The appearance of a para-complex and complex structure in the Euclidean and Minkowskian theory, respectively, is traced back to properties of the underlying R-symmetry groups. Finally, we indicate how our work will be extended to other types of multiplets and to supergravity in the future and explain the relevance of this project for the study of instantons, solitons and cosmological solutions in supergravity and M-theory.

hep-th↗

arXiv2002

Towards a third-order topological invariant for magnetic fields

An expression for a third-order link integral of three magnetic fields is presented. It is a topological invariant and therefore an invariant of ideal magnetohydrodynamics. The integral generalizes existing expressions for third-order invariants which are obtained from the Massey triple product, where the three fields are restricted to isolated flux tubes. The derivation and interpretation of the invariant shows a close relationship with the well-known magnetic helicity, which is a second-order topological invariant. Using gauge fields with an SU(2) symmetry, helicity and the new third-order invariant originate from the same identity, an identity which relates the second Chern class and the Chern-Simons three-form. We present an explicit example of three magnetic fields with non-disjunct support. These fields, derived from a vacuum Yang-Mills field with a non-vanishing winding number, possess a third-order linkage detected by our invariant.

physics.plasm-ph↗