David Demirdjian

360 Washington street, apt. #2

Somerville MA 02143

phone: (617) 953 1244

e-mail: demirdji@csail.mit.edu

French citizen, Green card holder

 

 

CURRICULUM VITAE

 

RESEARCH EXPERIENCE

 

2009 – present             Vecna Robotics, Cambridge, USA

Senior Research Scientist, R&D computer vision & autonomous robotics

¨      Research supervision, project management, grant proposal writing

¨      Projects:

·         Manipulation (object detection, 3D pose estimation)

·         Human-machine interaction (human pose estimation, gesture recognition)

·         Navigation (SLAM, visual odometry, scene recognition)

·         Sensor fusion (multimodal fusion, auto-calibration)

 

Additional affiliations:

-          MIT Media Lab (Prof. Cynthia Breazeal), project advisor in machine vision

-          MIT CSAIL (Prof. Randy Davis), research affiliate

 

2008 – 2009                Toyota Research Institute, Cambridge, USA

Principal Research Scientist, R&D automotive safety applications

¨      Research in computer vision and machine learning

¨      Project management (2 engineers), technical advisor

 

2002 – 2008                MIT Computer Science and Artificial Intelligence Laboratory, Cambridge, USA

Research scientist in computer vision, human-machine interaction (with Prof. Trevor Darrell)

¨      Research in real-time human pose estimation and gesture recognition: 

- Computer vision: markerless motion capture, tracking, object recognition and scene understanding, affective computing, multiview geometry, stereo

- Machine learning: temporal event classification, gesture recognition

- Multimodal interfaces (gesture & speech), ubiquitous computing

¨      Major supervised projects:

·         Interactive Wall: Real-time 3D gesture-speech interface system

·         CALO (Darpa): gaze and gesture analysis for meeting understanding

·         Spaulding: gait analysis for biomedical applications

·         Ford: facial expression recognition for communication-error detection

·         EStereo: real-time stereo C++/SSE library (open source, SourceForge)

 

Demonstrations can be viewed at: http://people.csail.mit.edu/demirdji/

 

2001 – 2009                Consulting services in real-time computer vision, human-machine interaction:  Newt Global (Irving, TX), Energid (Cambridge, MA) , Lensar (Orlando, FL), Mirclair (Boston, MA), Scalable Displays (Cambridge, MA)

2000 – 2002                MIT Artificial Intelligence Laboratory, Cambridge, USA

                                    Postdoctoral researcher in Vision Interface Group (Prof. Trevor Darrell)

Research and software development on smart environments and perceptual user interfaces                        

 

EDUCATION

1997 – 2000    Ph.D. in Computer Science at INRIA (Institut National de Recherche en Informatique et Automatisme), France                                                                                                Multiview geometry, stereo, autocalibration, motion segmentation                             Advisor: Prof. Radu Horaud

1994 – 1997    Engineering degree of ENSTA (Ecole Nationale Supérieure de Techniques Avancées),                              Paris, France. Majors: Compute Vision, Robotics and Artificial Intelligence  

REFERENCE

Prof. Trevor J. Darrell (trevor@eecs.berkeley.edu)    Associate Professor at UC Berkeley

Dr. Liu Qiao (liu.qiao@tema.toyota.com)                  General Manager, Technical Research Department, Toyota Research Institute.

 

Prof. Eric L. Grimson (welg@csail.mit.edu)              Head of the Department of Electrical Engineering and Computer Science, MIT

 

SKILLS

 

Languages: C/C++, Java, Matlab, MMX/SSE assembly

Programming: GPU (CUDA), Multithreading, ROS, OpenCV, OpenGL, Qt

Motion Capture Systems: VICON

Operating Systems: Linux, Windows, ROS

 

RESEARCH INTERESTS

 

Robotics: grasping and manipulation, situational awareness, SLAM, human-machine interfaces

Computer vision: 2D/3D object recognition and pose estimation, gesture recognition, affective computing, motion estimation, multi-view geometry, stereo and autocalibration.

 

 

SERVICE

 

Member of IEEE Boston section

Invited Panelist, National Science Foundation review panel

Program committee for the IEEE International Conference on Multimedia & Expo since 2007

Program committee for the ACM Multimedia 2007, 2008, 2009 conference

Program committee for the AAAI Symposium on Human Behavior Modeling 2009

Reviewer for CVPR, ICCV, ECCV since 2005

Reviewer for the ACM International Conference on Multimodal Interaction since 2005

Reviewer for 3DTV-CON 2009, Symposium on User Interface Software and Technology 2006, International Symposium on Wearable Computers 2007

Session chair at ICVS 2009, ICMI 2009

 

Frequent reviewer for major international journals:

IEEE Transactions on Pattern Analysis and Machine Intelligence, International Journal of Computer Vision, Computer Vision and Image Understanding, Journal of Computer Science and Technology, IEEE Transactions on Robotics, International Journal of Social Robotics, Intelligent Transport Systems

 

ADVISING

 

Current Master’s theses supervised

Yale Song, “A Framework for Recognizing Director Gestures on a Flight Carrier”, with Prof. R. Davis, 2010 (MIT, CSAIL)

 

PhD student supervised (MIT, CSAIL)

Sy Bor Wang, “Detecting Communication Errors using Facial Expressions”, with Prof. T. Darrell, 2008

 

Master’s theses supervised (MIT, CSAIL)

Chris Wilkens, “A Dynamic Key Frames Approach to Object Tracking”, 2007 (now PhD student, UCB)

Jasper Vicenti, “Aural Imaging from 3D Vision”, 2006 (now with iRobot)

Petch Manoharn, “Reasoning Agents for an Activity Recognition System”, 2005

Teresa Ko, “Untethered Human Motion Recognition for a Multimodal Interface,” with Prof. T. Darrell, 2003 (now PhD student, UCLA)

 

PUBLICATIONS

 

International conferences

· D. Demirdjian and C. Varri. Recognizing Events with Temporal Random Forests. Proceedings of the International Conference on Multimodal Interfaces, Cambridge, USA, 2009.

· D. Demirdjian and C. Varri. Driver Pose Estimation with a 3D Time-of-Flight Sensor. IEEE Workshop on Computational Intelligence in Vehicles and Vehicular Systems, Nashville, USA, 2009.

· D. Demirdjian and C. Varri. Recognizing Gestures for Virtual and Real World Interaction. Proceedings of the International Conference on Vision Systems, Belgium, 2009.

· D. Demirdjian and S. Wang. Recognition of Temporal Events using Multiscale Bags of Features. IEEE Workshop on Computational Intelligence for Visual Intelligence, 2009

· D. Demirdjian and R. Urtasun. Patch-based Pose Inference with a Mixture of Density Estimator. IEEE International Workshop on Analysis and Modeling of Faces and Gestures (AMFG), 2007.

· S. Wang, D. Demirdjian and T. Darrell. Detecting communication errors from visual cues during the system's conversational turn. Proceedings of the International Conference on Multimodal Interfaces, 2007.

· S. Wang, D. Demirdjian, H. Kjellstrom and T. Darrell. Multimodal Communication Error Detection for Driver-Car Interaction. In the 4th International Conference on Informatics in Control, Automation and Robotics, 2007.

· S. Wang, A. Quattoni, L. Morency, D. Demirdjian, T. Darrell, Hidden Conditional Random Fields for Gesture Recognition, Proceedings IEEE Conf. on Computer Vision and Pattern Recognition, 2006.

· L. Taycher, G. Shakhnarovich, D. Demirdjian, T. Darrell, Conditional Random People: Tracking Humans with CRFs and Grid Filters, Proc. IEEE Conference on Computer Vision and Pattern Recognition, 2006.

· D. Demirdjian, L. Taycher, G. Shakhnarovich, K. Grauman, T. Darrell, Avoiding the Streetlight Effect: Tracking by Exploring Likelihood Modes, Proceedings of the International Conference on Computer Vision, 2005.

· S. Wang, David Demirdjian, Inferring Body Pose using Speech Content, Proceedings of the International Conference on Multimodal Interfaces, 2005.

· D. Demirdjian, Combining Geometric and View-Based Approaches for Articulated Pose Estimation, Proceedings of the European Conference on Computer Vision, 2004.

· K. Tollmar, D. Demirdjian and T. Darrell. Navigating in Virtual Environments using a Vision-based Interface. In Proceedings of NordiCHI, Tampere, Finland, October 2004.

 

· D. Demirdjian. Enforcing Constraints for Human Body Tracking. In Workshop on Multi-Object Tracking, June 2003.

 

· K. Tollmar, D. Demirdjian and T. Darrell. Gesture + Play: Full Body Interaction for Virtual Environments. . In Proceedings of CHI’03, Fort Lauderdale, Florida, 2003.

· D. Demirdjian and T. Darrell. 3-D Articulated Pose Tracking for Untethered Deictic Reference. In Proc. ICMI’02, October 2002, Pittsburgh, Pennsylvania, 2002.

· D. Demirdjian, K. Tollmar, K. Koile, N. Checka and T. Darrell. Activity maps for location-aware computing. In Proc. IEEE Workshop on Applications of Computer Vision (WACV2002), Orlando, Florida, 2002.

· D. Demirdjian and T. Darrell. Motion Estimation from Disparity Images. In Proceedings of ICCV’01, Vancouver, Canada, July 2001, p. 213-218 , volume I, 2001.

 

· T. Darrell, D. Demirdjian, N. Checka and P. Felzenszwalb. Plan-view trajectory estimation with dense stereo background models. In Proceedings of ICCV’01, Vancouver, Canada, p. 628-635, 2001.

   

· D. Demirdjian, A. Zisserman and R. Horaud. Stereo Autocalibration from One Plane. In Proceedings of Sixth European Conference on Computer Vision, Dublin, Ireland, p. 625-639, volume II, 2000.

 

· D. Demirdjian and R. Horaud. A Projective Framework for Scene Segmentation in the Presence of Moving Objects. In Proceedings of IEEE CVPR'99, Fort Collins, Co, USA, p. 2-8, June 1999.
  

· D. Demirdjian, G. Csurka, and R. Horaud. Autocalibration in the presence of critical motions. Proceedings of the British Machine Vision Conference, Southampton, UK, p. 751-759, 1998.
 

· G. Csurka, D. Demirdjian, A. Ruf, and R. Horaud. Closed-form Solutions for the Euclidean Calibration of a Stereo Rig. In Proceedings of ECCV'98, Freiburg, Germany, p. 426-442, 1998.

 

Refereed journals articles

· D. Demirdjian, T. Ko, Trevor Darrel. Untethered Gesture Acquisition and Recognition for Virtual World Manipulation, Virtual Reality, 2005.

· D. Demirdjian and T. Darrell. Using Multiple-Hypothesis Disparity Maps and Image Velocity for 3D Motion Estimation. International Journal on Computer Vision, 2001.

· D. Demirdjian and R. Horaud. Motion-Egomotion Discrimination and Motion Segmentation from Image-pair Streams. In Computer Vision and Image Understanding, 78(1), p. 53-68, April 2000. 

 

· R. Horaud, G. Csurka, and D. Demirdjian. Stereo Calibration Using Rigid Motions,  IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(12), December 2000, p. 1446-1452.    

 

· G. Csurka, D. Demirdjian, and R. Horaud. Finding the Collineation between two Projective Reconstructions. Computer Vision and Image Understanding, 75(3), p. 260-269, September 1999.

 

PATENTS

 

·         Real-time stereo-based body tracking system, under MIT license.

·         5 Patents under submission to USPTO, in machine learning, computer vision

 

GRANTS

 

à  Dexterous object manipulation, with Vecna Technologies. STTR Darpa 2010, $100k.

à  Framework for pallet loading with a robotic arm, with Vecna Technologies. STTR Army 2010, $100k.

à  Anomalous activity recognition, with Vecna Technologies. STTR Air Force 2010, $90k.

à  Development of a markerless System for clinical analysis of gait abnormalities, with P. Bonato, CIMIT grant, 2006-2007, $50k

à  Expressive interaction for information and driving assistance, with T. Darrell, J. Glass. Ford, 2006-2007, $120k           

à  3-D tracking of human motion with stereo cameras, with T. Darrell, Toyota, 2005-2007, $190k

à  Modeling and tracking people using digital cameras  for virtual studio application, with C. Schmid, MIT-France Grant, 2003-2004.

à  Markerless gesture recognition system for unmanned air vehicle traffic control, with Energid. STTR, 2001.

 

INVITED TALKS

 

Imaging Technology for Driver Awareness (Executive Toyota workshop, Nagoya, Japan, 2009)

Pose Estimation using Local Features (International Workshop on Object Recognition, Italy, 2008)

Visual Perception (Toyota Motor Company, Cambridge, USA, 2007)

Autocalibration of Cameras in Meeting Environments (INRIA, France, 2006)

Stereo-based Articulated Tracking (Brown University, Providence, 2004)

Physical Awareness for Multimodal Human-Computer Interaction:

·         Boston University, Boston, 2003

·         NTT, Japan, 2003

·         ATR, Japan, 2003

Constraining Articulated Tracking (INRIA, France, 2002)

 

MISCELLENOUS

 

Open source software

Creator of EStereo, open-source C++/SSE library for real-time stereo map estimation.

(repository on sourceforge.net, ~300 download/month)

 

Community Service

  • Volunteer in local community (Boston Cares), for elderly and homeless
  • Tutoring in local schools (Somerville, MA)