Head Pose Estimation in Single- and Multi-view Environments - Results on the CLEAR'07 Benchmarks

Authors:
Michael Voit;Kai Nickel;Rainer Stiefelhagen
Affiliations:
Interactive Systems Lab, Universitt Karlsruhe (TH), Germany;Interactive Systems Lab, Universitt Karlsruhe (TH), Germany;Interactive Systems Lab, Universitt Karlsruhe (TH), Germany
Venue:
Multimodal Technologies for Perception of Humans
Year:
2008

Citing 5
Cited 2

Computing 3-D head orientation from a monocular image sequence

FG '96 Proceedings of the 2nd International Conference on Automatic Face and Gesture Recognition (FG '96)
A model-based gaze tracking system

IJSIS '96 Proceedings of the 1996 IEEE International Joint Symposia on Intelligence and Systems
Simultaneous Tracking of Head Poses in a Panoramic View

ICPR '00 Proceedings of the International Conference on Pattern Recognition - Volume 3
A Probabilistic Framework for Joint Head Tracking and Pose Estimation

ICPR '04 Proceedings of the Pattern Recognition, 17th International Conference on (ICPR'04) Volume 4 - Volume 04
Neural network-based head pose estimation and multi-view fusion

CLEAR'06 Proceedings of the 1st international evaluation conference on Classification of events, activities and relationships

Visual Focus of Attention in Dynamic Meeting Scenarios

MLMI '08 Proceedings of the 5th international workshop on Machine Learning for Multimodal Interaction
A System for Probabilistic Joint 3D Head Tracking and Pose Estimation in Low-Resolution, Multi-view Environments

ICVS '09 Proceedings of the 7th International Conference on Computer Vision Systems: Computer Vision Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we present our system used and evaluated on the CLEAR'07 benchmarks, both on single- and multi-view head pose estimation. The benchmarks show a high contrast in the application domain: whereas the single-view task provides meeting recordings involving high-quality captures of the participants, the multi-view benchmark targets at low-quality, unobtrusive observations of people by means of multiple cameras in an unconstrained scenario. We show that our system performs with state-of-the-art results under both conditions.