Virtual Representations and Modeling of Large-scale environments (VRML)


9:00 am - 10:10 am (chair: Frank Dellaert)
Improving Stereo Sub-Pixel Accuracy for Long Range Stereo, Stefan K. Gehrig and Uwe Franke
Evaluation of Large Scale Scene Reconstruction, Paul Merrell, Philippos Mordohai, Jan-Michael Frahm and Marc Pollefeys
10:40 am - 12:10 pm (chair: Jan-Michael Frahm)
Automatic Camera Network Localization using Object Image Tracks, Marci Meingast, Songhwai Oh, Shankar Sastry
A systematic approach for 2D-image to 3D-range registration in urban environments, Lingyun Liu and Ioannis Stamos
Towards Wiki based Dense City Modeling, Arnold Irschara, Christopher Zach and Horst Bischof
2:00 pm - 2:50 pm, invited talk (chair: Jan-Michael Frahm)
Modeling the World from Photos on the Internet, Steve Seitz
2:50 pm - 4:00 pm (chair: Frank Dellaert)
Simplifying the Reconstruction of 3D Models using Parameter Elimination, Daniel G. Aliaga, Ji Zhang, Mireille Boutin
3-D Reconstruction from Sparse Views using Monocular Vision, Ashutosh Saxena, Min Sun and Andrew Y. Ng

Call for papers

Recent developments in computer vision have improved the performance of structure from motion and multi-view stereo techniques towards the reconstruction of large scale environments. At the same time progress in computer graphics and network technology allow for effective visualization and delivery of such visual representations. As consumer applications are being deployed, a wide field of applications is opening up for real-world 3D modeling techniques.

Despite the enormous progress there are still many challenges left. Very large scale structure from motion with millions of images has open problems of: effectively scaling computations, consistency over large areas. Small artifacts often visually degrade relatively accurate reconstructions. How can one deal with windows or other surfaces with more complicated surface properties, trees and other occluding objects, which cause often holes in the reconstructions. The variability and change occurring in real-world environments poses additional challenges. The appearance of a scene varies over the course of the day and with the seasons, changes in geometry also occur at different temporal scales, from cars/pedestrians to building construction/destruction. Often, a complete detailed geometric reconstruction might not be the most appropriate model and image-based representations can be preferable for some applications.

The workshop invites researchers from computer vision, computer graphics, photogrammetry and related areas to present their work addressing the above problems:

Important dates:

Paper submission extended to: July 27nd, 2007
Notification August 27nd, 2007
Camera Ready August 31st, 2007
Workshop The workshop will be held on October 14th in Rio, Brazil.

Author information: (submission now open!)

Paper format: ICCV format
length: up to 8 pages (no summeray page)
submission page: via e-mail to: VRML.ICCV2007@gmail.com
submission information: No paper id required, please indicate double submission with ICCV and provide the paper number


Jan-Michael Frahm, UNC Chapel Hill, USA
Marc Pollefeys, UNC Chapel Hill, USA
Frank Dellaert, Gerogia Tech, USA
Jana Kosecka, George Mason U., USA

Program committee