Head Gesture Recognition in Intelligent Interfaces: The Role of Context in Improving Recognition

Author(s): Louis-Philippe Morency, Trevor Darrell
Venue: Proceedings of the 11th international conference on Intelligent user interfaces
Date: 2006

Type of Experiement: Survey/Multi-Case Study
Sample Size: 19


Location: Sydney, Australia
Pages: 32 - 38

The authors of this paper explore the use of head gestures for common user interface commands. They explore two different perceptual user interfaces: dialog box confirmation, and document browsing. The head gesture recognition system consists of a stereo-camera system. This is feed into a support vector machine that can recognize head gesture movement. The system also uses contextual clues to improve the accuracy and decrease the number of false positives in the system.

The authors perform a usability study that consists desktop computer setup with a stereo-camera mounted on the top of a 21" screen. The users would use the system to perform two tasks: reading a short text and answering three related questions. The tasks were conducted in three phases: conventional input only, head gesture input only, and user selected input. The authors found that head gestures was preferred more than conventional input in dialog box answering. However it was not preferred in document browsing. The authors also conducted a survey that rated the naturalness and efficiency of the interface methods that confirmed their results. There were 19 participants in the study. The authors also performed an analysis of the true positive and false positive rate of their system.