The Proteus Project

Proteus is a versatile, interactive 3D structure-from-motion implementation that allows a user to trace the contours of an object in a video and then attempts to extract a 3D model from it.

Abstract

Proteus is a semi-automatic structure-from-motion implementation that makes it straightforward – even for novice users – to obtain 3D models of real-world objects from digital video sequences. This is facilitated by the user tracing object outlines in key frames of the video.


The process of doing so is supported by a novel guidance technique that interactively "snaps" the annotations to object outlines. As a result, Proteus can be used for rapid interactive capture of 3D models that can be used in virtual and augmented reality scenarios such as computer games, online worlds and urban modelling.


Downloads

» Final Dissertation (PDF, 13 MB)
» Poster and abstract published at VMV 2009
» Synopsis for SET Award
» Talk given at the Women@CL lunch
» Project Proposal

N.B.: The Proteus project has now been completed and the dissertation about it has been submitted. The code is available for anyone to download and use, licensed under GPLv2.


Bibtex

  @inproceedings{ProteusInteractiveSFM,
          author    = {Malte Schwarzkopf and Christian Richardt},
          title     = {Proteus - Semi-Automatic Interactive Structure-from-Motion},
          booktitle = {Vision, Modeling and Visualization Workshop},
          url       = {http://proteus.malteschwarzkopf.de/},
          month     = november,
          year      = {2009},
          location  = {Braunschweig, Germany},
        }
  @misc{ProteusInteractiveSFMDiss,
          author    = {Malte Schwarzkopf},
          title     = {Proteus - Interactive Annotation-Based 3D Structure-from-Motion},
          url       = {http://proteus.malteschwarzkopf.de/},
          month     = may,
          year      = {2009},
          location  = {Cambridge, UK},
        }
"That's an insane project."
--- Russ Williams
All content © 2008 by Malte Schwarzkopf