New: CVPR 2024 – Perspective Autocalibration Solved


The Applied Algebra and Geometry group uses algebra, geometry, and optimization to address research questions in computer vision, robotics, and machine learning.

We are at the Czech Institute of Informatics, Robotics and Cybernetics at the Czech Technical University in Prague.

We are funded by

  • ROBOPROX – Robotics and advanced industrial production Project
  • A Unified 3D Map Representation GACS EXPRO 23-07973X  Project
  • ELIAS – European Lighthouse of AI for Sustainability – HORIZONT-CL4-2022-HUMAN-02 Project
  • ELISE – European Learning and Intelligent Systems Excellence – H2020-ICT-48 Project
  • RICAIP – European Learning and Intelligent Systems Excellence – ESIF OP VV
  • SPRING – Socially Pertinent Robots in Gerontological Healthcare – EU H2020 
  • MAGIK Eye – CTU Laboratory

News

Perspective Autocalibaton Solved @ CVPR 2024
A Porfiri Dal Cin, T Duff, L Magri, T Pajdla. Minimal Perspective AutocalibrationCVPR 2024

ICCV 2023 Best Paper Finalist
F Arrigoni, T Pajdla, A Fusiello. Viewing Graph Solvability in Practice.

AAG organizes
3DV 2022 – International Conference in 3D Vision in Prague

CVPR 2022 Best Paper
P Hruby, T Duff, A Leykin, T Pajdla. Learning to solve hard minimal problems.

ICCV 2021 Best Paper Honorable Mention
F Arrigoni, A Fusiello, E Ricci, T Pajdla. Viewing graph solvability via cycle consistency

AAG co-organizes ECCV 2020
Long-Term Visual Localization under Changing Conditions
Visual Localization estimates the position and orientation of a camera from images.

AAG co-organizes the CVPR 2020
Joint Workshop on Long-Term Visual Localization, Visual Odometry and Geometric and Learning-based SLAM 
Visual Localization estimates the position and orientation of a camera from images. Simultaneous Localization and Mapping (SLAM) tracks camera motion while simultaneously building a (3D) map of the scene. 

Best Student Paper at ICCV 2019

T Duff, K Kohn, A Leykin, T Pajdla. PLMP-Point-Line Minimal Problems in Complete Multi-View Visibility. ICCV 2019 [pdf,supp,code]

AAG co-organizes ICCV 2019
Workshop on 3D Reconstruction in the Wild

We are interested in 3D reconstruction techniques capable of operating in unconstrained conditions which are robust and/or real-time including multi-camera calibration, feature extraction, matching, data fusion, depth learning, and meshing techniques to 3D modeling approaches.

AAG co-organizes CVPR 2019
Tutorial on The Art of Solving Minimal Problems

Minimal solvers play an important role in many computer vision problems such as 3D Reconstruction, Visual Localization, Augmented/Mixed Reality, Visual Odometry or Robotics. The goal of the tutorial is to explain the principles behind solving minimal problems and give practical means for engineers and researchers to be able apply the recent methods.

AAG co-organizes CVPR 2019
Workshop on Autonomous Driving

The CVPR 2019 Workshop on Autonomous Driving (WAD) gathers researchers and engineers from academia and industry to discuss the latest advances in perception for autonomous driving, arguably the most promising application of computer vision and AI in general.

AAG co-organizes CVPR 2019
Workshop on Long-Term Visual Localization under Changing Conditions

Visual localization is the problem of (accurately) estimating the position and orientation, i.e., the camera pose, from which an image was taken with respect to some scene representation. It is a vital component in many interesting Computer Vision and Robotics scenarios, including autonomous vehicles such as self-driving cars and other robots, Augmented/Mixed/Virtual Reality, Structure-from-Motion, and SLAM.

ARTWIN Solutions for Industry and construction 4.0
EU H-2020 Project (Partners: Q-PLAN, B<>COM, Siemens, CTU in Prague, Nokia Bell Labs, Holo-Light, Artefacto)

Project signed. Kick-off on 1 Oct 2019.

ACCV 2018 Best Paper
J Pritts, Z Kukelova, V Larsson, O Chum.
Rectificaƒtion from Radially-Distorted Scales.
ACCV 2018

6 Dec 2018 – AAG Participated at the creation of ELLIS Society – European Laboratory for Learning and Intelligent Systems

 

Recent Results  

 


T Duff, V Korotynskiy, T Pajdla, M Regan. Using monodromy to recover symmetries of polynomial systems. ISSAC 2023 | pdf


A Lasaruk, T Pajdla. An efficient model for a camera behind a parallel refractive slab. International Journal of Computer Vision 2023 | pdf


F Arrigoni, T Pajdla, A Fusiello. Viewing Graph Solvability in Practice. ICCV 2023 The Best Paper Finalist | pdf


P Hruby, V Korotynskiy, T Duff, L Oeding, M Pollefeys, T Pajdla, V Larsson. Four-view geometry with unknown radial distortion. CVPR 2023 | pdf | supp | github


T Duff, V Korotynskiy, T Pajdla, M H Regan. Galois/monodromy groups for decomposing minimal problems in 3D reconstruction. SIAM Journal on Applied Algebra and Geometry 2022 | arXiv


R Fabbri, T Duff, H Fan, M Regan, D da Costa de Pinho, E Tsigaridas, C Wampler, J Hauenstein, P J Giblin, B B Kimia, A Leykin, T Pajdla. Trifocal relative pose from lines at points. IEEE TPAMI 2022 | pdf


A Adam, T Sattler, K Karantzalos, T Pajdla. Objects can move: 3D change detection by geometric transformation consistency. ECCV 2022 | arXiv


F Arrigoni, A Fusiello, R Rizzi, E Ricci, T Pajdla. Revisiting viewing graph solvability: an effective approach based on cycle consistency. IEEE TPAMI 2022 | pdf


M Dubenova, A Zderadickova, O Kafka, T Pajdla, M Polic. D-InLoc++: Indoor Localization in Dynamic Environments. DAGM German Conference on Pattern Recognition 2022 | arXiv


T Kato, I Shimizu, T Pajdla. Improving image pair selection for large scale Structure from Motion by introducing modified Simpson coefficient. IEICE TRANSACTIONS on Information and Systems 2022 | pdf


P Trutman, M Safey El Din, D Henrion, T Pajdla. Globally optimal solution to inverse kinematics of 7DOF serial manipulator. IEEE Robotics and Automation Letters 2022 | pdf | arXiv | ICRA 2023 (oral)


F Arrigoni, E Ricci, T Pajdla. Multi-frame motion segmentation by combining two-frame results.
International Journal of Computer Vision 2022 | pdf


K Stepanova, J Rozlivek, F Puciow, P Krsek, T Pajdla, M Hoffmann. Automatic self-contained calibration of an industrial dual-arm robot with cameras using self-contact, planar constraints, and self-observation. Robotics and Computer-Integrated Manufacturing 2022 | pdf


E Martyushev, J Vrablikova, T Pajdla. Optimizing elimination templates by greedy parameter search. CVPR 2022 | pdf | supp | github


P Hruby, T Duff, A Leykin, T Pajdla. Learning to solve hard minimal problems. CVPR 2022 (oral) The Best Paper. pdf | supp | arXiv | github


F Arrigoni, L Magri, T Pajdla. Motion segmentation with pairwise matches and unknown number of motions. ICPR 2021 | pdf


F Arrigoni, A Fusiello, E Ricci, T Pajdla. Viewing graph solvability via cycle consistency.  ICCV 2021 (oralThe Best Paper Honorable Mention  | pdf | supp


C Toft, W Maddern, A Torii, L Hammarstrand, E Stenborg, D Safari, M Okutomi, M Pollefeys, J Sivic, T Pajdla, F Kahl, T Sattler. Long-Term Visual Localization Revisited. IEEE TPAMI 2020 | pdf 


Z Kukelova, C. Albl, A Sugimoto, T Pajdla, K Schindler. Minimal Rolling Shutter Absolute Pose with Unknown Focal Length and Radial Distortion. ECCV 2020 (Spotlight) [pdf] [supp


T Duff, K Kohn, A Leykin, T Pajdla. PL1P – Point-line Minimal Problems under Partial Visibility in Three Views. ECCV 2020 [pdf] [supp]


F Arrigoni, L Magri, T Pajdla. On the Usage of the Trifocal Tensor in Motion Segmentation. ECCV 2020 [pdf] [supp]


D Barath, M Polic,  Z Kukelova, T Sattler, W Forstner, T Pajdla. Making Affine Correspondences Work in Camera Geometry Computation. ECCV 2020 [pdf] [supp]


D Zhukov, J-B Alayrac, I Laptev, J Sivic. Learning actionness via long-range temporal order verification. ECCV 2020 [pdf[supp]


Y Labbe, J Sivic, J Carpentier, M Aubry. Multi-view multi-object 6D pose estimation via robust scene consistency optimization. 
ECCV 2020 [pdf[supp]


I Rocco, J Sivic, R Arandjelovic. 
Efficient Neighbourhood Consensus Networks via Submanifold Sparse Convolutions. ECCV 2020


I Rocco, M Cimpoi, R Arandjelovic, A Torii, T Pajdla, J Sivic. NCNet: Neighbourhood Consensus Networks for Estimating Image Correspondences. IEEE PAMI Early Access, 2020 [pdf


C Albl, Z Kukelova, V Larsson, T Pajdla. Rolling Shutter Camera Absolute Pose. IEEE PAMI 42:(6) 1439-1452, 2020 [pdf]


C Albl, Z Kukelova, V Larsson, M Polic, T Pajdla, K Schindler. From Two Rolling Shutters to One Global Shutter. CVPR 2020 (ORAL)  [pdf] [supp] [arXiv


A Miech, J-B Alayrac, L Smaira, I Laptev, J Sivic, A Zisserman. End-to-End Learning of Visual Representations From Uncurated Instructional Videos. CVPR 2020 (ORAL) [pdf] [arXiv] [video]


M Polic, S Steidl, C Albl, Z Kukelova, T Pajdla. Uncertainty Based Camera Model Selection. CVPR 2020 [pdf] [supp]


R Fabbri, T Duff, H Fan, M H Regan, D da Costa de Pinho, E Tsigaridas, C W Wampler, J D Hauenstein, P J Giblin, B Kimia, A Leykin, T Pajdla. TRPLP – Trifocal Relative Pose From Lines at Points. CVPR 2020 [pdf] [supp


T Duff, K Kohn, A Leykin, T Pajdla. PLMP – Point-Line Minimal Problems in Complete Multi-View Visibility. ICCV 2019 | pdf (ORAL & ICCV 2019 The Best Student Paper Award)