Call For Papers

Call For Papers

The deadline has now long passed, see the schedule for the list of accepted papers.

Significant progress has been made over the past two decades in recovering the 3D structure from a set of images, for both static and dynamic scenes. However, the focus has been mostly placed on matching primitive tokens (lines, edges, and regions), followed by recovery of corresponding 3D primitives without regard to their relationships or context in the scene. Recent approaches have been proposed to exploit intermediate to higher-level knowledge. Intermediate knowledge includes meshing, surfaces, ordering, occlusion, and parallelism and orthogonality of structures, while higher-level information involves recognition of scene elements and architectural scenes. In motion tracking, parametric models of human motion and/or 3D avatars have been used to enable robust motion recovery, and parse human motion at a higher level of description.

This workshop is aimed towards bringing together computer vision researchers who are pursuing this area of research, in the hope to facilitate discussions and debates that will identify the opportunities and challenges in applying prior knowledge to the still unsolved problem of 3D structure recovery from images. Relevant topics for the workshop include but are not limited to:

  • prior knowledge in architecture
  • parameterized models for human motion (including face modeling and articulated motion), hand, body, and face tracking using learning
  • automatic single image modeling
  • recovering surface meshes from unstructured point clouds
  • learning exemplars
  • reasoning about occlusion
  • role of object recognition in SFM
  • Bayesian model selection for structure recovery


  • Paper submission: June 30 2003
  • Notification of acceptance: Aug 21
  • Camera-ready copy: Sept 1

Papers should be submitted electronically in Acrobat pdf format with no more than 8 pages including all figures, references and appendices using the ICCV 2003 submission format. Submissions should be emailed to Frank Dellaert, dellaert@cc.gatech.edu.

Please also include in your email:

  • Paper title
  • Abstract
  • Keywords
  • Author name(s) and affiliation(s)
  • Name, address, email, phone and fax number for the contact author

[HLK 2003] [Call For Papers] [Schedule]