3D Tracking = Classification + Interpolation

  • Authors:
  • Carlo Tomasi;Slav Petrov;Arvind Sastry

  • Affiliations:
  • -;-;-

  • Venue:
  • ICCV '03 Proceedings of the Ninth IEEE International Conference on Computer Vision - Volume 2
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

Hand gestures are examples of fast and complex motions.Computers fail to track these in fast video, but sleight ofhand fools humans as well: what happens too quickly wejust cannot see. We show a 3D tracker for these types ofmotions that relies on the recognition of familiar configurationsin 2D images (classification), and fills the gapsin-between (interpolation). We illustrate this idea with experimentson hand motions similar to finger spelling. Thepenalty for a recognition failure is often small: if two configurationsare confused, they are often similar to eachother, and the illusion works well enough, for instance, todrive a graphics animation of the moving hand. We contributeadvances in both feature design and classifier training:our image features are invariant to image scale, translation,and rotation, and we propose a classification methodthat combines VQPCA with discrimination trees.