Optimisation Methods in
Geometric Vision

27-31 January 2019, Shonan Village Centre


Group photo - Thank you for the wonderful time! See you again in the future!


  • Thanks to Laurent Kneip and Kenichi Kanatani for these wonderful shots.
  • Click on photos for larger version.
  • Not all meeting participants are in the photos.


Presentation slides, summary of meeting outcomes.



A warm welcome from the organisers!

Tat-Jun Chin Anders Eriksson Yasuyuki Matsushita


We thank our generous sponsors



What is a Shonan Meeting?


Excerpt from NII:


NII Shonan Meetings, following the well-known Dagstuhl Seminars, aim to internationally promote informatics its research, by providing another world's premier venue for world-class scientists, promising young researchers, and practitioners to come together in Asia to exchange their knowledge, discuss their research findings, and explore a cutting-edge informatics topics.


The meetings are held in Shonan Village Center near Tokyo, which offers facilities for conferences, trainings, and lodging in a resort-like setting. The friendly and open atmosphere is to promote communications among participants. The NII Shonan Meetings are managed by National Institute of Informatics (NII), Japan.


Meeting Abstract


Computer vision is concerned with inferring properties of the world from observations in the form of visual images. Such inverse problems typically take shape as optimization problems, that aim to find the best explanation, for the complex visual phenomenon that gave rise to a set of noisy and incomplete visual measurements. For computer vision applications to be successful, the underlying optimization problems must be supported by efficient and dependable solution methods.


The proposed meeting focuses on a broad subclass of computer vision problems called geometric vision problems. Roughly, these are problems that exploit fundamental geometrical constraints arising from the image formation process or physical properties of the scene (e.g., lighting conditions, characteristics of motions), to extract information of the scene (e.g., depth, 3D shape, camera trajectory, object identities) from the given visual data. Example geometric vision problems include structure-from-motion (SfM), simultaneous localization and mapping (SLAM), pose averaging, photometric stereo, and motion segmentation. Methods for solving geometric vision problems underpin many useful applications, such as 3D reconstruction, robot navigation, object recognition/tracking, and computational photography.


Geometric vision is replete with hard optimization problems. By "hard", we mean that the time needed to solve the optimization problems grows quickly with the size of the input data. Take, for example, the task of robustly estimating the planar perspective transformation (a.k.a. homography) from outlier-contaminated point correspondences between two images. Due to the inherent intractability of robust homography estimation, practitioners often rely on simple randomized heuristics to find rough approximate solutions, which neither guarantee optimality nor provide bounds on the approximation error.


The computational difficulty of geometric vision problems is also often compounded by the extremely large size of the input. Take, for example, the task of bundle adjustment, i.e., calculate 3D points and camera poses that are consistent with a set of images of a scene. In the age of big data, the input image set is often obtained by "scraping" Internet photo collections, or by conducting long-term surveillance of a scene using a robot. Such input sizes easily overwhelm traditional computing architectures, and distributed or parallel versions of bundle adjustment must be used.


Technical Themes


The overall theme for the proposed meeting is recent theoretical and algorithmic advances on optimization problems in geometric vision. These include, but are not restricted to:

  • Solvability and approximability of geometric vision problems.
  • Duality in geometric vision problems.
  • Global optimization algorithms for geometric vision.
  • Approximate algorithms including randomized methods.
  • Distributed and incremental algorithms for geometric vision problems.
  • Machine learning and deep learning in geometric vision problems.
Following the spirit of Shonan Meetings, we will consider other related topics, based on the interest of the attendees and the trajectory of the discussions.


It is hoped that the meeting will contribute towards promoting the value of basic algorithmic research in computer vision, especially in the area of geometric optimization. From a more practical standpoint, and as alluded to above, geometric vision problems underpin some of the most important capabilities (e.g., 3D sensing and navigation, object recognition and tracking) that support intelligent machines. Therefore, the outcomes of the proposed meeting could have significant impact on some of the most important technological developments, such as self-driving cars, intelligent domestic robots, and smart manufacturing - developments that are already driving the economic growth in developed and developing countries alike.


Keynote Speakers

Fredrik Kahl Frank Dellaert Richard Hartley



Alessio Del BueItalian Institute of Technology
Alexander BronsteinTechnion, Israel Institute of Technology
Ali Bab-HadiasharRMIT University
Anders ErikssonQueensland University of Technology
Carl OlssonChalmers Institute of Technology
Christopher ZachChalmers University of Technology
Danda Pani PaudelETH Zurich
Daniel Lin Wen YanOsaka University
David SuterEdith Cowan University
Florian BernardMax-Planck-Institute for Informatics
Frank DellaertGeorgia Tech
Fredrik KahlChalmers University of Technology
Gim Hee LeeNational University of Singapore
Hongdong LiAustralian National University
Jamie SherrahDigital Animal Interactive Inc. (FTSY)
Jesus BrialesFacebook
Kenichi KanataniOkayama University
Laurent KneipShanghai Tech
Luca CarloneMIT
Michael BrownYork University
Michael WaechterOsaka University
Michael BronsteinUniversity of Italian Switzerland, Intel, Imperial
Ping TanSimon Fraser University
Richard HartleyAustralian National University
Robert MahonyAustralian National University
Simon LuceyCarnegie Mellon University
Sudipta SinhaMicrosoft Research
Tarek HamelUniversite Cote d'Azur
Tat-Jun ChinThe University of Adelaide
Viorela IlaUniversity of Sydney
Yasuyuki MatsushitaOsaka university
Yinqiang ZhengNII
Yongduek SeoSogang University




Contact the organisers if there are any issues with the program.


Sunday 27 January

1500 onwardsCheck-in
1900-2100Welcome banquet


Monday 28 January

0900-0915Welcoming address - Tat-Jun Chin, Anders Eriksson, Yasuyuki Matsushita
0915-1000Keynote 1 - Fredrik Kahl
1000-1015Michael Brown
1015-1030Christopher Zach
1030-1100Coffee break
1100-1115Robert Mahony
1115-1130Yongduek Seo
1130-1200Breakout session 1
1330-1345Sudipta Sinha
1345-1400Yinqiang Zheng
1400-1415Michael Waechter
1415-1430Viorela Ila
1430-1500Breakout session 2
1500-1530Coffee break
1530-1545Hongdong Li
1545-1600David Suter
1600-1630Breakout session 3
1630-1800Free time


Tuesday 29 January

0900-0915Program briefing
0915-1000Keynote 2 - Frank Dellaert
1000-1015Ping Tan
1015-1030Jamie Sherrah
1030-1100Cofee break
1100-1115Ali Bab-Hadiashar
1115-1130Gim Hee Lee
1130-1200Breakout session 4
1315-1330Group photo
1330-1345Kenichi Kanatani
1345-1400Jesus Briales
1400-1415Carl Olsson
1415-1445Breakout session 5
1445-1515Coffee break
1515-1530Florian Bernard
1530-1545Alessio Del Bue
1545-1600Danda Pani Paudel
1600-1630Breakout session 6
1630-1800Free time


Wednesday 30 January

0900-0915Program briefing
0915-1000Keynote 3 - Richard Hartley
1000-1015Tarek Hamel
1015-1030Laurent Kneip
1030-1100Coffee break
1100-1115Simon Lucey
1115-1130Luca Carlone
1130-1200Breakout session 7
1300-2045Excursion and dinner


Thursday 31 January

0730-0930Breakfast and check-out
0930-0945Yasuyuki Matsushita
0945-1000Anders Eriksson
1000-1015Tat-Jun Chin
1015-1100Coffee break
1100-1130Breakout session 8
1130-1200Conclusion and wrap-up
1200-1330Lunch and end of meeting