no code implementations • 19 Feb 2024 • Jannik Brinkmann, Abhay Sheshadri, Victor Levoso, Paul Swoboda, Christian Bartelt
We anticipate that the motifs we identified in our synthetic setting can provide valuable insights into the broader operating principles of transformers and thus provide a basis for understanding more complex models.
1 code implementation • 12 Oct 2023 • Paul Roetzer, Ahmed Abbas, Dongliang Cao, Florian Bernard, Paul Swoboda
In this work we propose to combine the advantages of learning-based and combinatorial formalisms for 3D shape matching.
no code implementations • ICCV 2023 • Jannik Brinkmann, Paul Swoboda, Christian Bartelt
Therefore, we measure the impact of training data, model architecture, and training objectives on social biases in the learned representations of ViTs.
1 code implementation • NeurIPS 2023 • Duy M. H. Nguyen, Hoang Nguyen, Nghiem T. Diep, Tan N. Pham, Tri Cao, Binh T. Nguyen, Paul Swoboda, Nhat Ho, Shadi Albarqouni, Pengtao Xie, Daniel Sonntag, Mathias Niepert
While pre-trained deep networks on ImageNet and vision-language foundation models trained on web-scale data are prevailing approaches, their effectiveness on medical tasks is limited due to the significant domain shift between natural and medical images.
no code implementations • 28 Jan 2023 • Ahmed Abbas, Paul Swoboda
We propose a graph clustering formulation based on multicut (a. k. a.
no code implementations • 4 Dec 2022 • Duy M. H. Nguyen, Hoang Nguyen, Mai T. N. Truong, Tri Cao, Binh T. Nguyen, Nhat Ho, Paul Swoboda, Shadi Albarqouni, Pengtao Xie, Daniel Sonntag
Recent breakthroughs in self-supervised learning (SSL) offer the ability to overcome the lack of labeled training samples by learning feature representations from unlabeled data.
1 code implementation • 1 Jul 2022 • Stefan Haller, Lorenz Feineis, Lisa Hutschenreiter, Florian Bernard, Carsten Rother, Dagmar Kainmüller, Paul Swoboda, Bogdan Savchynskyy
To address these shortcomings, we present a comparative study of graph matching algorithms.
1 code implementation • 23 May 2022 • Ahmed Abbas, Paul Swoboda
Our solver achieves significantly faster performance and better dual objectives than its non-learned version, achieving close to optimal objective values of LP relaxations of very large structured prediction problems and on selected combinatorial ones.
1 code implementation • CVPR 2022 • Paul Roetzer, Paul Swoboda, Daniel Cremers, Florian Bernard
We present a scalable combinatorial algorithm for globally optimizing over the space of geometrically consistent mappings between 3D shapes.
no code implementations • 4 Feb 2022 • Paul Swoboda, Bjoern Andres, Andrea Hornakova, Florian Bernard, Jannik Irmai, Paul Roetzer, Bogdan Savchynskyy, David Stein, Ahmed Abbas
In order to facilitate algorithm development for their numerical solution, we collect in one place a large number of datasets in easy to read formats for a diverse set of problem classes.
no code implementations • CVPR 2022 • Duy M. H. Nguyen, Roberto Henschel, Bodo Rosenhahn, Daniel Sonntag, Paul Swoboda
Multi-Camera Multi-Object Tracking is currently drawing attention in the computer vision field due to its superior performance in real-world applications such as video surveillance in crowded scenes or in wide spaces.
1 code implementation • CVPR 2022 • Ahmed Abbas, Paul Swoboda
We present a massively parallel Lagrange decomposition method for solving 0--1 integer linear programs occurring in structured prediction.
1 code implementation • CVPR 2022 • Ahmed Abbas, Paul Swoboda
We propose a highly parallel primal-dual algorithm for the multicut (a. k. a.
2 code implementations • ICCV 2021 • Andrea Hornakova, Timo Kaiser, Paul Swoboda, Michal Rolinek, Bodo Rosenhahn, Roberto Henschel
We present an efficient approximate message passing solver for the lifted disjoint paths problem (LDP), a natural but NP-hard model for multiple object tracking (MOT).
2 code implementations • NeurIPS 2021 • Ahmed Abbas, Paul Swoboda
We propose a fully differentiable architecture for simultaneous semantic and instance segmentation (a. k. a.
Ranked #8 on Panoptic Segmentation on Cityscapes test
1 code implementation • ICML 2020 • Andrea Hornakova, Roberto Henschel, Bodo Rosenhahn, Paul Swoboda
We present an extension to the disjoint paths problem in which additional \emph{lifted} edges are introduced to provide path connectivity priors.
Ranked #2 on Multi-Object Tracking on 2D MOT 2015
1 code implementation • 14 Apr 2020 • Stefan Haller, Mangal Prakash, Lisa Hutschenreiter, Tobias Pietzsch, Carsten Rother, Florian Jug, Paul Swoboda, Bogdan Savchynskyy
We demonstrate the efficacy of our method on real-world tracking problems.
1 code implementation • 14 Apr 2020 • Stefan Haller, Paul Swoboda, Bogdan Savchynskyy
This property allows to significantly reduce the computational time of the combinatorial solver and therefore solve problems which were out of reach before.
5 code implementations • 25 Mar 2020 • Michal Rolínek, Paul Swoboda, Dominik Zietlow, Anselm Paulus, Vít Musil, Georg Martius
Building on recent progress at the intersection of combinatorial optimization and deep learning, we propose an end-to-end trainable architecture for deep graph matching that contains unmodified combinatorial solvers.
Ranked #3 on Graph Matching on PASCAL VOC
no code implementations • CVPR 2019 • Paul Swoboda, Dagmar Kainm"uller, Ashkan Mokarian, Christian Theobalt, Florian Bernard
We present a convex relaxation for the multi-graph matching problem.
1 code implementation • ICCV 2019 • Ahmed Abbas, Paul Swoboda
We consider general discrete Markov Random Fields(MRFs) with additional bottleneck potentials which penalize the maximum (instead of the sum) over local potential value taken by the MRF-assignment.
no code implementations • 26 Nov 2018 • Florian Bernard, Johan Thunberg, Paul Swoboda, Christian Theobalt
The matching of multiple objects (e. g. shapes or images) is a fundamental problem in vision and graphics.
1 code implementation • CVPR 2019 • Paul Swoboda, Vladimir Kolmogorov
We present a new proximal bundle method for Maximum-A-Posteriori (MAP) inference in structured energy minimization problems.
no code implementations • CVPR 2017 • Paul Swoboda, Bjoern Andres
We propose a dual decomposition and linear program relaxation of the NP -hard minimum cost multicut problem.
1 code implementation • CVPR 2017 • Paul Swoboda, Carsten Rother, Hassan Abu Alhaija, Dagmar Kainmueller, Bogdan Savchynskyy
We study the quadratic assignment problem, in computer vision also known as graph matching.
1 code implementation • CVPR 2017 • Paul Swoboda, Jan Kuske, Bogdan Savchynskyy
We propose a general dual ascent framework for Lagrangean decomposition of combinatorial problems.
no code implementations • 9 Jan 2016 • Jörg Hendrik Kappes, Paul Swoboda, Bogdan Savchynskyy, Tamir Hazan, Christoph Schnörr
We present a probabilistic graphical model formulation for the graph clustering problem.
no code implementations • CVPR 2015 • Alexander Shekhovtsov, Paul Swoboda, Bogdan Savchynskyy
We propose an efficient implementation, which runs in time comparable to a single run of a suboptimal dual solver.
no code implementations • CVPR 2014 • Paul Swoboda, Alexander Shekhovtsov, Jörg Hendrik Kappes, Christoph Schnörr, Bogdan Savchynskyy
We propose a novel polynomial time algorithm to obtain a part of its optimal non-relaxed integral solution.
no code implementations • NeurIPS 2013 • Bogdan Savchynskyy, Jörg Hendrik Kappes, Paul Swoboda, Christoph Schnörr
We consider energy minimization for undirected graphical models, also known as MAP-inference problem for Markov random fields.
no code implementations • 16 Jan 2013 • Paul Swoboda, Christoph Schnörr
We present a novel variational approach to image restoration (e. g., denoising, inpainting, labeling) that enables to complement established variational approaches with a histogram-based prior enforcing closeness of the solution to some given empirical measure.