Speaker: Mark Holler, Intel Microprocessor Research Lab
Title: Digital Cubicle: Maximizing Visual Interaction within an Office Cubicle

Abstract: Business office productivity has reached a plateau with the current 2D windows interface and single monitor display. Limited display requires substantial time for non-value adding manipulation of windows. 3D graphical user interfaces have not been able to provide a net productivity improvement due to additional complexity in navigating and the same problems associated with limited display area. Advances in projection technology and support for multiple displays in today’s operating systems provide an opportunity to address the basic problem of limited display area. Projected displays also enable new physical configurations which invite multi-modal interaction. This talk presents one such configuration of 3 projectors in a standard office cubicle which we refer to as the Digital Cubicle. Relative to a conventional 20″ monitor It provides double the number of pixels and more than twice the field of view at a more comfortable viewing distance of18-48″. In the Digital Cubicle we have the ability to capture and display human faces at full scale at distances typical of interpersonal interactions. The front projected displays allow a miniature camera to be placed behind a hole in the screen to capture frontal face images with apparent eye contact. The 90 degree format of the screens provides a collaborative space where gaze can be used by both participants to focus attention on shared objects. Results of initial experiences using 2D and 3D user interfaces in and arm pointing gesture tracking in this space for general business tasks and two person collaborations will be described.

Speaker Background: Mark Holler leads the Visual Interactivity group within Intel’s Microprocessor Research Lab in Santa Clara, Ca. Since 1996 his technical focus has been on computer vision for human computer interaction. The group develops components for face, hand, and body gesture recognition and tracking which are being distributed in Intel’s Computer Vision Library. In the last year Mark’s research interests have expanded to include complete multimodal systems where computer vision components can showcase their ability to improve productivity. Prior to his work in computer vision he managed implementation of optimized Image Processing, Signal Processing and Pattern Recognition Libraries for the Pentium Processor with MMX(TM) technology. Mark was also responsible for Intel’s Neural Network program which produced the 80170NX and NI1000 Neural Network chips and associated tools

The seminar is free. Please join our speaker for dinner ($10-20) afterwards. Directions to the restaurant will be handed out at the seminar.

Location: SEMI Technology Building, 815 E. Middlefield Road, Mountain View, CA