Panel #3: Better Frameworks for Face and Gesture Recognition

Chair & Moderator: Peter Stucki / University of Zurich

Panelists: TBA

Wednesday October 16, 1996, 10:15am- 11:15noon

Summary

In the field of information technology, new technical developments occur with enormous rapidity. Yet a fast development phase is most likely followed by a maturing phase in which the pace of technical development slows down. This is the time to begin to think and initiate streamlining activities in order to put forward a coherent set of process and interface specifications that may lead to an open system. The fields of face and gesture recognition have now reached this phase: while a wide range of approaches and implementations are available as separate entities, their interworking could dramatically enhance overall applicability [1]. Streamlining and formalizing aim to define and ensure performance attributes and features of products and systems, as well as precise methods of testing and measuring their attributes and features. They can reduce uncertainties and thus foster the development of a market place for new and emerging technologies such as face and gesture recognition. In this context, a formalism may be defined as a specification approved by a recognized body of experts. It takes the form of a document created with the involvement and agreement of many interested parties. The harmonization of a variety of proposals and drafts to define an open system is generally a long and tedious process and in practice, this task is by far not trivial. Therefore, it is proposed to first conceive and create a framework environment in which individual, partial or complete draft proposals may be implemented and integrated for rapid application development and testing [2].

Typically, such framework environments build upon architectural specifications that interlink functional specifications and definitions. Ideally, frameworks should provide a programming environment, guarantee portability, hardware and I/O independence, and allow the rapid prototyping of specific pilot applications. It is expected that the experience gained and the results obtained will lead to an open systems prototype that, after a maturing process, may provide useful inputs for the definition of possible standards at a later point in time. The purpose of this panel is to discuss the potential for speci=1Ffying and formalizing specific tasks in the field of face and gesture recognition at large and to possibly initiate two discussion forums. The first, called the Face Recognition Expert Group (FREG), will be chartered to define a recognition framework architecture and to select and discuss robust algorithms and procedures for experimentation. The second, called the Gesture Recognition Expert Group (GREG) will do so for gesture recognition related aspects.

If successful, expert group findings and recommendations shall be discussed in a forthcoming workshop on face and gesture recognition.

References:

[1] Proceedings of the International Workshop on Automatic Face- and Gesture-Recognition, M. Bichsel, Ed., June 26-28, 1995, Zurich, Switzerland.
[2] Ackermann Ph., Developing Object-Oriented Multimedia Software Based on the MET++ Application Framework, dpunkt, Heidelberg, 1996 (ISBN3-920993-52-7).

2d International Conference on Automatic Face- and Gesture-Recognition , fg96@media.mit.edu

Last modified: Tue Sep 10 09:52:57 EDT 1996