Dr. Kevin Köser
Research Group Leader
Oceanic Machine Vision Group
RD2 / Marine Geosystems
Phone: +49 431 600- 2566
Kevin Köser is the head of the research group Oceanic Machine Vision at the GEOMAR Helmholtz Centre for Ocean Research, Kiel. His main research interest lies in 3D underwater robot vision and novel, automated, camera-based measurement techniques for (deep) sea environments and processes (3D underwater machine vision). These help to explore, map and monitor (deep) sea habitats, to study resources, or to assess hazards. This involves physical models of underwater light transport and imaging, as well as models on how marine quantities can be observed. Based on such insights, novel, automated vision methods (inspired by human vision abilities with respect to distances, sizes, semantics) are sought that allow to quantify and measure different environments in 2D, 3D or 4D. Here, 3D mapping and reconstruction from deep sea photos is one of the big challenges.
Before joining GEOMAR Dr. Köser has taught the classes 3D Photography and the Computer Vision Lab at the Swiss Federal Institute of Technology (ETH Zurich) and has worked as a senior researcher in ETH's Computer Vision and Geometry Lab on shape and motion extraction from photos and videos, camera (auto-) calibration as well as geolocalization and image registration.
Services for the Community
He is a program committee member for the major international computer vision conferences such as ECCV, ICCV, CVPR, ... and serves as a reviewer for several journals in machine vision and photogrammetry. He is also the co-organizer of the 2021 ICCV-Workshop for Computer Vision in the Ocean. His research has been awarded several prizes, such as a Prof. Petersen Prize for Technology (VDI/VDE/PWP Foundation), a dissertation award (University of Kiel) and the DAGM 2011 Main Prize (German Association for Pattern Recognition). In 2019, he has been selected for the Emmy-Noether Programme of the German Research Foundation.
In the context of 3D underwater computer vision (automatic image- and video-based quantification, classification, 3D reconstruction, mapping, ...) there are opportunities for computer science master students for a thesis or student jobs at GEOMAR. We devise and program automated approaches for visual and acoustic processing, which are then applied both in shallow water (e.g. -40m in the North Sea) as well as in the deep sea (e.g. -5000m in the Pacific Ocean), and for some spots nobody has ever been there before. Possible tasks and topics include robust estimation, efficient large scale numerics, GPU computation, differential raytracing, 3D and 4D geometry, image processing, underwater navigation and geodata processing, robotics, helping with constructing sensor systems for deep sea deployment consisting of cameras and/or sonar, small computers, storage devices and other sensors or actuators. Please check the Oceanic Machine Vision homepage for opportunities.
For a more exhaustive list, see google scholar .
She M., Song Y., Mohrmann J., Köser K.: Adjustment and Calibration of Dome Port Camera Systems for Underwater Vision. In: Fink G., Frintrop S., Jiang X. (eds) Pattern Recognition. DAGM GCPR 2019. Lecture Notes in Computer Science, vol 11824. Springer, Cham (preprint pdf)
Song, Y. , Köser, K., Kwasnitschka, T., Koch, R. Iterative Refinement for Underwater 3d Reconstruction: Application to Disposed Underwater Munitions in the Baltic Sea ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. XLII-2/W10, 2019, pp.181-187,
Hongbo Liu, Jan Sticklus, Kevin Köser, Henk-Jan T. Hoving, Hong Song, Ying Chen, Jens Greinert, and Timm Schoening:
J. Quintana, R. Garcia, R. Campos, L.Neumann, T. Weiss, K. Köser, J. Mohrmann, J. Greinert: Towards automatic recognition of mining targets using an autonomous robot. MTS/IEEE OCEANS 2018, Charleston, USA.
Timm Schoeing, Kevin Köser, Jens Greinert:
Anne Jordt, Kevin Köser, Reinhard Koch:
T. Kwasnitschka, K.Köser, J.Sticklus, M.Rothenbeck, T.Weiß, E.Wenzlaff, T.Schoening, L. Triebe, A. Steinführer, C.Devey, J.Greinert:
Durden, Jennifer M., Schoening, Timm, Althaus, Franziska, Friedman, Ariell, Garcia, Rafael, Glover, Adrian G., Greinert, Jens, Jacobsen Stout, Nancy, Jones, Daniel O.B., Jordt, Anne, Kaeli, Jeffrey, Köser, Kevin, Kuhnz, Linda A., Lindsay, Dhugal, Morris, Kirsty J., Nattkemper, Tim W., Osterloff, Jonas, Ruhl, Henry A., Singh, Hanumant, Tran, Maggie and Bett, Brian J.:
A.Jordt, C.Zelenka, J. Schneider v. Deimling, R.Koch, K.Köser:
Anne Sedlazeck, Kevin Koeser, and Reinhard Koch Supporting Underwater Archaeology by 3D Reconstruction from Underwater Images. Skyllis Journal of Underwater Archaeology, vol. 10 (2), 2010
|Anne Sedlazeck, Kevin Koeser and Reinhard Koch 3D Reconstruction Based on Underwater Video from ROV Kiel 6000 Considering Underwater Imaging Conditions. IEEE OCEANS Conference 2009, Bremen, Germany, 2009 |
Camera Calibration and Self-Calibration
B. Li, L. Heng, K. Köser, M. Pollefeys: A Multiple-Camera System Calibration Toolbox Using a Feature Descriptor-Based Calibration Pattern. IEEE/RSJ Int. Conf. on Intelligent Robots and Systems (IROS), 2013.
J. Brito, R. Angst, K. Köser, M. Pollefeys: Radial Distortion Self-Calibration. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2013,
J. Brito, R.Angst, K.Köser,C. Zach, P.Branco, M. Ferreira, M. Pollefeys: Unknown Radial Distortion Centers in Multiple View Geometry Problems. Asian Conference of Computer Vision (ACCV 2012), Daejeon, Korea, pp.136-149
Kevin Koeser, Christian Beder and Reinhard Koch : Conjugate Rotation: Parameterization and Estimation from an Affine Feature Correspondence. IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2008). Anchorage, June 2008
3D Registration and 3D Feature Representation
B. Zeisl, K. Köser, M. Pollefeys: Automatic Registration of RGB-D Scans via Salient Directions. Computer Vision (ICCV), 2013 International Conference on, pp.2808-2815, 1-8 Dec. 2013,
B. Zeisl, K.Köser, M.Pollefeys: Viewpoint Invariant Matching via Developable Surfaces. European Conference on Computer Vision 2012, WS on Consumer Depth Cameras, Florence, Italy, 2012 Lecture Notes in Computer Science Volume 7584, pp 62-71
|G. Baatz, O. Saurer, K. Köser, M. Pollefeys Leveraging Topographic Maps for Image to Terrain Alignment. Joint 3DIM/3DPVT Conference: 3D Imaging, Modeling, Processing, Visualization & Transmission (3DIMPVT 2012), Zurich, 2012|
|Kevin Koeser and Reinhard Koch Exploiting Uncertainty Propagation in Gradient-Based Image Registration. British Machine Vision Conference (BMVC 2008) Leeds, UK, 2008 BibTeX |
|Kevin Koeser and Reinhard Koch: Perspectively Invariant Normal Features. Proceedings of IEEE 11th International Conference on Computer Vision 2007 (3D Representation for Recognition Workshop), Rio de Janeiro, Brazil, 14-21 Oct. 2007|
|Kevin Koeser, Christian Perwass and Gerald Sommer: Dense Optic Flow with a Bayesian Occlusion Model. LNCS 3667 (Spatial Coherence in Visual Motion Analysis), pp. 127-139, Springer Verlag, Berlin, Heidelberg, 2006 |
Stereo, 3D Reconstruction, Structure from Motion / SLAM
O. Saurer, K. Köser, J.-Y. Bouguet, M. Pollefeys: Rolling Shutter Stereo. Computer Vision (ICCV), 2013 International Conference on, pp.465-472, 1-8 Dec. 2013,
B.Jacquet, C. Häne, K.Köser, M.Pollefeys: Normal Map Capture for Nearly Flat Reflective Surfaces. Computer Vision (ICCV), 2013 International Conference on, pp.713-720, 1-8 Dec. 2013,
|K. Köser, C. Zach, M. Pollefeys Dense 3D Reconstruction of Symmetric Scenes From a Single Image. Lecture Notes in Computer Science 6835 (DAGM 2011), Springer Verlag |
|Bogumil Bartczak, Kevin Koeser, Felix Woelk and Reinhard Koch: Extraction of 3D Freeform Surfaces as Visual Landmarks for Real-Time Tracking. Journal of Real-time Image Processing 2007(2):81-101, Springer-Verlag, Berlin, Heidelberg The original publication is available at springerlink.com. |
Space Resection / Camera Tracking / Pose Estimation
|Kevin Koeser and Reinhard Koch Differential Spatial Resection - Pose Estimation Using a Single Local Image Feature. European Conference on Computer Vision (ECCV 2008) Marseille, France, 2008 BibTex Poster The original publication is available at springerlink.com.|
|Kevin Koeser, Christian Beder and Reinhard Koch : Conjugate Rotation: Parameterization and Estimation from an Affine Feature Correspondence. IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2008). Anchorage, June 2008|
|Kevin Koeser, Bogumil Bartczak and Reinhard Koch: Robust GPU-Assisted Camera Tracking using Free-form Surface Models. Journal of Real-time Image Processing 2007(2):133-147, Springer-Verlag, Berlin, Heidelberg BibTeX The original publication is available at springerlink.com. |
|Kevin Koeser, Bogumil Bartczak and Reinhard Koch: An Analysis-by-Synthesis Camera Tracking Approach based on Free-form Surfaces. Lecture Notes in Computer Science 4713 (DAGM 2007), pp. 122–131, Heidelberg, 2007|
|G. Thomas, J. Chandaria, B. Bartczak, K. Koeser, R. Koch, M. Becker, G. Bleser, D. Stricker, C. Wohlleber, M. Felsberg, F. Gustafsson, J. D. Hol, T. B. Schön, J. Skoglund, P. J. Slycke, S. Smeitz: Realtime Camera Tracking in the MATRIS Project. Motion Imaging Journal 2007, vol. 116, no. 7-8, p. 266-271, Society of the Motion Picture and Television Engineers, London, 2007|
|Kevin Koeser, Bogumil Bartczak and Reinhard Koch: Drift-free Pose Estimation with Hemispherical Cameras. European Conference for Visual Media Production (CVMP 2006), pp. 20-28, November 2006, London |
|Jan-Michael Frahm, Kevin Köser and Reinhard Koch: Pose Estimation for Multi-Camera Systems. LNCS 3175 (26th DAGM Symposium, Tübingen, Germany), pp. 27-35, Springer-Verlag |
Geolocalization and Landmark Recognition
|G. Baatz, O. Saurer, K. Köser, M. Pollefeys: Leveraging Topographic Maps for Image to Terrain Alignment. Joint 3DIM/3DPVT Conference: 3D Imaging, Modeling, Processing, Visualization & Transmission (3DIMPVT 2012), Zurich, 2012|
|G. Baatz, O. Saurer, K. Köser, M. Pollefeys: Large Scale Visual Geo-Localization of Images in Mountainous Terrain. European Conference on Computer Vision (ECCV 2012), Florence, Italy, 2012 |
|G. Baatz, K. Köser, D. Chen, R. Grzeszczuk, M. Pollefeys: Leveraging 3D City Models for Rotation Invariant Place-of-Interest Recognition. International Journal of Computer Vision, Vol. 96, Issue 3 (2012) (Special Issue on Mobile Vision), pp. 315-334,|
|D. Chen, G. Baatz, K. Koeser, S. Tsai, R. Vedantham, T. Pylvanainen, K. Roimela, X. Chen, J. Bach, M. Pollefeys, B. Girod, R. Grzeszczuk: City-Scale Landmark Identification on Mobile Devices. IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2011), Colorado Springs, USA |
|Georges Baatz, Kevin Koeser, David Chen, Radek Grzeszczuk and Marc Pollefeys: Handling Urban Location Recognition as a 2D Homothetic Problem Lecture Notes in Computer Science 6316 (ECCV 2010), pp. 266-279, Springer, 2010 |