HCI09 Program

9.30 - 11.05    Oral Session I: Face Analysis and Gesture Recognition

Incorporating Facial Features into a Multi-Channel Gesture Recognition System for the Interpretation of Irish Sign Language Sequences
   Daniel Kelly, Jane Reilly Delannoy, John Mc Donald and Charles Markham

Feature Selection of Facial Displays for Detection of Non Verbal Communication in Natural Conversation
   Tim Sheerman-Chase, Eng-Jon Ong and Richard Bowden

Automatic Detection of Facial Actions from 3D Data
Arman Savran and Bulent Sankur

Automatic Sign Segmentation from Continuous Signing via Multiple Sequence Alignment
   Pinar Santemiz, Oya Aran, Murat Saraclar and Lale Akarun

11.05 – 11.30   Coffee Break

11.30 – 12.30   Keynote presentation: Understanding How People Communicate With Each Other: Multimodal Approach for Modeling and Recognizing Multiparty Face-to-Face Conversations.

Kazuhiro Otsuka, NTT Corporation, Japan

This talk overviews the research trends and progress in communication scene analysis, especially that made in NTT CSL(CommunicationScience Labs.).The goal is to develop techniques and systems that can automatically understand how people communicate with each other through the exchange of nonverbal messages in face-to-face situations. To that end, we have focused on and developed multimodal approaches for modeling and recognizing multiparty meetings (more than two participants) by exploiting multimodal data from cameras and microphones. This talk introduces our works from the following perspectives. First, I will present a probabilistic framework that can model the relationship between the structure of conversation and peoples’ behaviors including utterances, head pose, gaze direction, and head gestures. Second, for tracking head pose, we have developed a novel particle filter called Memory-Based particle filter (M-PF); it keeps in memory past state estimates, and predicts prior distributions by temporal sampling according to the recurrent probability of the past states in the future. Its advantages include robustness against complex motion, and quick recovery from track loss. Moreover, I will describe a real-time system for meeting analysis that combines speaker diarization and face pose tracking using an omnidirectional camera and microphone array. Finally, I will describe recent progress on the automatic inference of interpersonal emotions that evolve during meetings; the cue is directed smiling, i.e. “who is smiling at whom”.

12.30 - 14.00   Lunch Break

14.00 - 15.35   Oral Session II: Applications

Learning to Rank Images from Eye Movements
Kitsuchart Pasupa, Craig J. Saunders, Sandor Szedmak, Arto Klami, Samuel Kaski and Steve R. Gunn

The Use of Tongue Protrusion Gestures for Video-based Communication
   Luis Ricardo Sapaico, Hamid Laga and Masayuki Nakajima

Object Recognition in Service Robots: Conducting Verbal Interaction on Color and Spatial Relationship
   Yoshinori Kuno, Katsutoshi Sakata and Yoshinori Kobayashi

Is Gender Recognition Affected by Age?
Guodong Guo, Charles Dyer, Yun Fu and Thomas Huang

15.35 – 16.00   Coffee Break

16.00 - 17.15     Oral Session III: Tracking

Tracking Humans Interacting with the Environment Using Efficient Hierarchical Sampling and Layered Observation Models
  Jan Bandouch and Michael Beetz

Shoulder Gesture Interface for Operating Electric Wheelchair
   Nobuhiko Sato, Ikushi Yoda and Takenobu Inoue

Real-Time Motion Control Using Pose Space Probability Density Estimation
Dumebi Okwechime and Richard Bowden

17.15 - 18.00   Panel