CVPR 2020 tutorial : Computational Geometry tools for Computer Vision
The goal of this tutorial is to introduce Computational Geometry tools and highlight their potential in Computer Vision.
Computational Geometry is a branch of Computer Science devoted to the study of geometric algorithms and data structures, has been used successfully in many fields including CAD, Computer Vision and Graphics.
Mature C++ libraries such as the Open Source Computational Geometry Algorithms Library (CGAL) and OpenMesh provide geometric data structures and algorithms that are reliable, efficient and easy to integrate in existing software.
- 1st hour : Introduction to the Open Source CGAL library by
Dr Andreas Fabri : slides (pdf).
- What is CGAL? (15 min)
- Demo for point set processing (15 min)
- Demo for polygon mesh processing (15 min)
- Demo for mesh generation (15 min), then a 15-min break
- 2nd hour : Graph reconstructions and cloud segmentations by
Dr Vitaliy Kurlin.
- Demo for hole detection and cloud segmentation by using CGAL (30 min)
- Demo for graph reconstructions by using CGAL (30 min), then a 15-min break
- 3rd hour : Polygonal meshes for resolution-independent superpixels by
Dr Andrew Fitzgibbon.
- Resolution-independent framework for superpixel over-segmentation (30 min)
- Demo for OpenMesh tools to handle polygonal meshes of superpixels (30 min)
- Short bios of the authors : Dr Andreas Fabri, Dr Vitaliy Kurlin, Dr Andrew Fitzgibbon.
Back to Top of this page | Back to Home page
Overview of 1-hour sessions with slides and references
The proposed educational course will comprehensively review the Computational Geometry tools that complement and leverage the OpenCV library. The most relevant and important applications for the CVPR community are superpixel segmentations and graph reconstructions. We aim the proposed talks below at graduate students studying computer vision. The mixture of the speakers from industry and academia will attract researchers from both worlds.
The first talk covers the Open Source Computational Geometry Algorithms Library in breadth and will include interactive demos for the novices. The second talk demonstrates CGAL tools for graph reconstructions [1,2], segmentations of images [3] and point clouds [4,5] with theoretical guarantees. The third talk will explain how the Open Source CGAL and OpenMesh [6] have helped to develop a new resolution-independent framework [7,8,9] for superpixel partitions into polygons whose vertices can have any real coordinates at subpixel resolution.
- [1] V. Kurlin. A one-dimensional homologically persistent skeleton of a point cloud in any metric space. Computer Graphics Forum, v. 34, p. 253-262, 2015.
- [2] V. Kurlin. A homologically persistent skeleton is a fast and robust descriptor of interest points in 2d images. In Proceedings of CAIP 2015, Lecture Notes in Computer Science,v. 9256, p. 606-617, 2015.
- [3] V. Kurlin and D. Harvey. Superpixels optimized by color and shape. In Proceedings of EMMCVPR 2017, Lecture Notes in Computer Science, v. 10746, p. 297-311, 2018.
- [4] V. Kurlin. A fast and robust algorithm to count topologically persistent holes in noisy clouds. In Proceedings of CVPR 2014, p. 1458-1463.
- [5] V. Kurlin. A fast persistence-based segmentation of noisy 2d clouds with provable guarantees. Pattern Recognition Letters, 83:3-12, 2016.
- [6] M. Botsch, S. Steinberg, S. Bischoff, L. Kobbelt. Openmesh - a generic and efficient polygon mesh data structure.
- [7] F. Viola, A. Fitzgibbon, and R. Cipolla. A unifying resolution-independent formulation for early vision. In Proceedings of CVPR, pages 494-501, 2012.
- [8] J. MacCormick and A. Fitzgibbon. Curvature regularization for resolution-independent images. In Proceedings of EMMCVPR, pages 165-179, 2013.
- [9] J. Forsythe, V. Kurlin, and A. Fitzgibbon. Resolution-independent superpixels based on convex constrained meshes. In LNCS (Proc. of ISVC), volume 10072, pages 223-233, 2016.
Past tutorials on CGAL were delivered at Siggraph (2008, 2016), Siggraph Asia (2009), EECV (2016). In the past 3 years there were no similar tutorials on Computational Geometry at top venues in Computer Vision and Graphics. The slides by Andreas Fabri are already online. The slides for other parts will be available by 18th May 2020.
Back to Top of this page | Back to Home page
Short bios of authors : Dr Andreas Fabri, Dr Vitaliy Kurlin, Dr Andrew Fitzgibbon
Dr Andreas Fabri is the CEO at the GeometryFactory, which he founded in 2003 to make industrially applicable the technology accumulated through European projects developing the Open Source Computational Geometry Algorithms Library (CGAL). Andreas Fabri obtained his PhD in Computational Geometry in 1994 from Ecole de Mines de Paris while working at Inria. He was among the first developers of the CGAL, which was initially funded in 1996 by the European Research Council. | |
Dr Vitaliy Kurlin is an Associate Professor in Computer Science at the University of Liverpool. He was awarded the Marie Curie International Incoming Fellowship (2005-2007) and the EPSRC grant Persistent Topological Structures in Noisy Images (2011-2013). In 2014-2016 he has gained industrial experience through Knowledge Transfer Secondments in the Computer Vision group at Microsoft Research, Cambridge, UK. Since 2018 he leads the Topological Data Analysis group at the Materials Innovation Factory and the Liverpool team of four co-Is funded by the £3.5M EPSRC 5-year grant Application-Driven Topological Data Analysis with the Universities of Oxford and Swansea. | |
Dr Andrew Fitzgibbon leads the All Data AI group at Microsoft Research in Cambridge, UK. He is best known for his work on 3D vision, having been a core contributor to the Emmy-award-winning 3D camera tracker “boujou“, to body tracking for Kinect for Xbox 360, and for the articulated hand-tracking interface to Microsoft’s HoloLens. He has published numerous highly-cited papers and received ten “best paper” prizes, the Silver medal of the Royal Academy of Engineering, and the BCS Roger Needham award. He is a fellow of the Royal Academy of Engineering, the British Computer Society, and the International Association for Pattern Recognition, and is a BMVA Distinguished Fellow. Before joining Microsoft in 2005, he was a Royal Society University Research Fellow at Oxford University, having previously studied at Edinburgh University, Heriot-Watt University, and University College, Cork. |
Back to Top of this page | Back to Home page