Data-Driven Shape Anlaysis
Course info
Course: CS468, Spring 2014
Location: Gates B12
Time: Tuesday, Thursday. 9:30-10:45
Instructors:
• Vladimir (Vova) Kim, vk2@stanford.edu
Office Hours: Tue, 11am-12pm
• Qixing (Peter) Huang, huangqx@stanford.edu
Office Hours: Fri, 1pm-2pm
Description:
With recent developments in 3D modeling software and advances in acquisition techniques for 3D geometry, the shape of a large numbers of objects has been digitized. Existing datasets include millions of real-world every day objects, cultural heritage artifacts, buildings, as well as medical, scientific, and engineering models. In this course we will discuss (1) computational tools for organizing 3D model collections, (2) algorithms for understanding their semantic properties and relations, and (3) applications in graphics, vision, and other areas. In particular, some of the topics that will be covered in this class include: reconstruction and recognition of 3D data, shape retrieval, segmentation and matching, probabilistic and grammar models for objects and scenes, and data-driven 3D vision. We will also cover machine learning and optimization techniques that are widely used in data-driven shape analysis.
Prerequisites:
Background assumed includes basic material in computer graphics, linear algebra, and optimization.
Grading (tentative):
| | Homeworks (3 assignments) | 60% |
| | Final Project | 40% |
Course Schedule
| Date |
Topic |
Materials |
Instructors |
Announcements |
| April 1 |
1. Introduction. Shape Descriptors |
Slides:
intro, SD
Covered Papers:
Shape Distributions,
Spin Images,
Lightfield
|
Vova |
|
| April 3 |
2. Shape Retrieval. Intrinsic Descriptors |
Slides: ICP, Apps, Intrinsic Descriptors
Shape Matching:
ICP
Intrinsic:
Shape Google,
GPS,
HKS,
WKS
Retrieval & Exploration:
Sketch and text,
Sketch only,
Quartets,
Templated Exploration
|
Vova |
Assignment 1 due Apr 20 |
| April 8 |
3. Pair-wise rigid registration |
Slides: RigidMatching |
Peter |
|
| April 10 |
4. Pair-wise non-rigid registration |
Slides: NonRigidRegistration |
Peter |
|
| April 15 |
5. Pair-wise intrinsic shape matching |
Slides: Intrinsic Methods |
Vova |
|
| April 17 |
6. Joint shape matching I |
Slides: JointMatchingI |
Peter |
|
| April 22 |
7. Joint shape matching II |
Slides: JointMatchingII |
Peter |
Assignment 2 due May 4th |
| April 24 |
8. Human-centric shape analysis |
Slides: Affordances |
Vova |
|
| April 29 |
9. Shape segmentation |
Slides: Segmentation |
Peter |
|
| May 1 |
10. Joint shape segmentation |
Slides: JointSegmentation |
Peter |
|
| May 6 |
11. Project proposals |
|
Peter and Vova |
|
| May 8 |
12. Shape classification |
Slides: Classification |
Peter |
|
| May 13 |
13. Parametric Shape Spaces / Shape Grammars |
Slides: Parametric Models |
Vova |
Assignment 3 due May 20th |
| May 15 |
14. Data-driven modeling |
Slides: Modeling |
Vova |
|
| May 20 |
15. Data-driven reconstruction |
Slides: Reconstruction |
Peter |
|
| May 22 |
16. City-scale reconstruction |
|
Guest lecturer: Qian-Yi Zhou |
|
| May 27 |
17. Data-driven scene analysis and synthesis |
Slides:Scene Analysis |
Vova |
|
| May 29 |
18. Data-driven techniques in 3D vision I |
Slides: 3DVision |
Peter |
|
| June 3 |
19. Wrap up |
Slides: WrapUp |
Peter |
|
| June 5 |
20. Final project presentations |
|
Peter and Vova |
|
References
April 8th
1. Random Sample Consensus: A Paradigm for Model Fitting with Applications to Image Analysis and Automated Cartography, M. Fischler and R. Bolles, June 1981, Comm. of the ACM
2. Linear Model Hashing and Batch RANSAC for Rapid and Accurate Object Recognition, Y. Shan, B. Matei, H. S. Sawhney, R. Kumar, D. Huber, and M. Hebert, CVPR 2004
3. A Spectral Technique for Correspondence Problems using Pairwise Constraints, M. Leordeanu and M. Hebert, ICCV 2005
4. Robust Global Registration, N. Gelfand, N. Mitra, L. Guibas and H. Pottmann, Symposium on Geometry Processing, 2005
5. Partial and Approximate Symmetry Detection for 3D Geometry, N. Mitra, L. Guibas, M. Pauly, SIGGRAPH 2006
6. Reassembling Fractures Objects by Geometric Matching, Q. Huang, S. Flory, N. Gelfand, M. Hofer and H. Pottmann, SIGGRAPH 2006
7. 4-points Congruent Sets for Robust Surface Registration, Dror Aiger, Niloy J. Mitra, Daniel Cohen-Or, SIGGRAPH 2008
April 10th
1. Articulated Object Reconstruction and Markerless Motion Capture from Depth Video, Y. Pekelny and C. Gotsman, Eurographics' 08
2. Global correspondence optimization for non-rigid registration of depth scans, H. Li, R. Sumner, M. Pauly, SGP’08
3. Non-Rigid Registration Under Isometric Deformations, Q. Huang, B. Adams, M. Wiche, and L. Guibas, SGP’08
4. Robust single-View geometry and motion reconstruction, H. Li, B. Adams, L. Guibas, M. Pauly, SIGGRAPH ASIA'09
5. 3D self-portraits, H. Li, E. Vouga, A. Gudym, J. Barron, L. Luo, G. Gusev, SIGGRAPH ASIA'13
6. Animation Cartography - Intrinsic Reconstruction of Shape and Motion, A. Tevs, A. Berner, M. Wand, I. Ihrke, M. Bokeloh, J. Kerber, H. Seidel, TOG'13
7. Dynamic Geometry Processing, W. Chang, H. Li, N. Mitra, M. Pauly, M. Wand: Eurographics 2012 tutorial
April 17th
1. Automatic Three-dimensional Modeling from Reality, PhD thesis, D. Huber, Robotics Institute, Carnegie Mellon University, 2002
2. Reassembling fractured objects by geometric matching. Q. Huang, S. Flory, N. Gelfand, M. Hofer, and H. Pottmann, SIGGRAPH’06
3. Disambiguating visual relations using loop constraints, C. Zach, M. Klopscjotz, and M. POLLEFEYS, CVPR’10
4. An optimization approach to improving collections of shape maps, A. Nguyen, M. Ben-Chen, K. Welnicka, Y. Ye, and L. Guibas, SGP ’11
5. Exploring collections of 3d models using fuzzy correspondences, V. G. Kim, W. Li, N. Mitra, S. Diverdi, and T. Funkhouser, SIGGRAPH’12
6. An Optimization Approach for Extracting and Encoding Consistent Maps in a Shape Collection, Q. Huang, G. Zhang, L. Gao, S. Hu, A. Bustcher, L. Guibas, SIGGRAPH ASIA’12
7. Consistent Shape Maps via Semidefinite Programming, Q. Huang and L. Guibas, SGP’13
8. Matching Partially Similar Objects via Matrix Completion, Y. Chen, L. Guibas and Q. Huang, ICML’14
May 22th
1. Elastic Fragments for Dense Scene Reconstruction, Q.-Y. Zhou, S. Miller and V. Koltun, ICCV 2013
2. Dense Scene Reconstruction with Points of Interest, Q.-Y. Zhou and V. Koltun, SIGGRAPH 2013
3. 2.5D Dual Contouring: A Robust Approach to Creating Building Models from Aerial LiDAR Point Clouds, Q.-Y. Zhou and U. Neumann, ECCV 2010
4. A Survey of Urban Reconstruction, P. Musialski, P. Wonka, D. Aliaga, M. Wimmer, L. Gool, W. Purgathofer, EUROGRAPHICS 2012
Acknowledgements
The course staff would like to thank the Stanford Computer Forum for their support.