S-Seer: Guiding Perceptual Sensing and Analysis with Value of Information
Abstract
In this project, we explore the use of Expected Value of Information (EVI) to control the use and analysis of data from multiple perceptual sensors in the Seer system for identifying office activities. Seer uses Layered Hidden Markov Models (LHMMs) at different temporal granularities for diagnosing office situations from real-time streams of evidence (video, audio, and computer interactions). We review the overall Seer architecture, describe how we integrated EVI analyses, and show how EVI computations endow Seer's descendant — S-Seer — with the ability to balance computation required for perceptual analysis with the discriminatory power of the sensors.
Publications
Selective Perception Policies for Guiding Sensing and Computation in Multimodal Systems: A Comparative Analysis Nuria Oliver and Eric Horvitz. Computer Vision and Image Understanding (CVIU).
Layered Representations for Learning and Inferring Office Activity from Multiple Sensory Channels Nuria Oliver, Ashutosh Garg and Eric Horvitz. Computer Vision and Image Understanding (CVIU).
Selective Perception Policies for Guiding Sensing and Computation in Multimodal Systems: A Comparative Analysis Nuria Oliver and Eric Horvitz. ICMI 2003, Vancouver, BC, November 2003.
Layered Representations for Human Activity Recognition Nuria Oliver, Eric Horvitz and Ashutosh Garg. ICMI 2002, Pittsburgh, October 2002.
Paper presented at CVPR 2001 (Cues in Communication Workshop) Nuria Oliver, Eric Horvitz and Ashutosh Garg.