Ent?te

Logo du LIFL

Depuis le 1er janvier 2015 le LIFL et le LAGIS forment le laboratoire CRIStAL

  1. Actualités

Séminaire de

B. Prabhakaran

18 juin 2010
Salle du conseil IRCICA

Multi-modal Sensing of Human Interaction

Sensing human intentions has been of great interest to research community typically using single modality such as video or audio. The accuracy of techniques working on single modality have been considerably lower. With recent advances in other sensing technologies, researchers have been combining modalities for better accuracies in understanding human intentions. For instance, a simple use of location-aware GPS has been able to provide better results in different applications.

In this talk, I give an overview of multimedia research in the University of Texas at Dallas in different areas such as human activity recognition, tele-immersion, physical medicine and rehabilitation. Specifically, I describe our results on a new approach to train Hidden Markov Models (for human action recognition) using a combination of modalities such as body sensors and vision data. We show that we can achieve very fast training times and also higher accuracies in recognizing human motions.

Ours

UMR 8022 - Laboratoire d'Informatique Fondamentale de Lille - Copyright © 2012 Sophie TISON - Crédits & Mentions légales

Page respectant XHTML et CSS.

Pour tout commentaire / Comments and remarks : webmaster