MultiCam - Cognitive Multicamera System Funded by the The Czech Academy of Sciences under Project 1ET101210407. Project start: 07/2004, expected end: 12/2008 Tomas Svoboda Project leader, multicamera (self)-calibration, articulated models Petr Doubek Multicamera appearance based tracking, background modeling Karel Zimmermann Advanced tracking methods, learning of articulated models of unknown complexity Pavol Vlcek Head detection and tracking

Project description

In this project a set of theoretically well established methods for modeling and recognition of events will be developed. The system will use multiple cameras that will observe scene from different perspectives. It will be based on consumer HW and essentially transportable and flexible. Newly developed event representation will alow user-defined expected event to be recognized. Moreover, with long-term installation, the system will learn knowledge about usual events autonomously and will recognize an unusual event then. Understanding of the dynamic events allows development of the VirtualEditor that will be switching between cameras depending on the scene context. A multicamera system with above specified functionality may be applied in homeland security applications, video surveillance, tele-presence, tele-teaching, human-machine interfacing and others.

Misc

• Demos. Images, Videos, etc. Student demos.
  
• Data. Three sequences of (almost) planar objects with hand-clicked ground truth.
• student works (mostly in Czech). [ Diploma | Bachelor theses | individual projects ]
  Complete prints: [ Diploma | Bachelor theses | individual projects ]
• Multi-Camera (Self)Calibration. A software package (in Matlab) for a complete geometrical calibration of multicamera systems.
  
• Internal: [ How-to capture | MultiCam Software ]

Selected Publications

• K. Zimmermann, T. Svoboda, J. Matas. Anytime learning for the NoSLLiP tracker. Image and Vision Computing. Accepted, pre-published on line doi:10.1016/j.physletb.2003.10.07, [pdf]

• K. Zimmermann, J. Matas, and T. Svoboda. Tracking by an Optimal Sequence of Linear Predictors. IEEE Transactions on Pattern Analysis and Machine Intelligence. 31(4), 2009, [pdf]

• K. Zimmermann, T. Svoboda, J. Matas. Simultaneous learning of motion and appearance. The 1st International Workshop on Machine Learning for Vision-based Motion Analysis, In conjunction with the 10th European Conference on Computer Vision 2008, [ paper | talk ].

• K. Zimmernann, T. Svoboda, J. Matas. Adaptive Parameter Optimization for Real-time Tracking. In Proceedings of the Workshop on Non-rigid Registration and Tracking through Learning - NRTL 2007 (joint workshop with the ICCV 2007). [ paper ]

• J. Matas, K.Zimmermann, T. Svoboda, A. Hilton. Learning Efficient Linear Predictors for Motion Estimation. In Proceedings of 5th Indian Conference on Computer Vision, Graphics and Image Processing. [ paper | talk ]

• K.Zimmermann, T. Svoboda, J. Matas Multiview 3D Tracking with an Incrementally Constructed 3D Model, Third International Symposium on 3D Data Processing, Visualization and Transmission (3DPVT), Chapel Hill, USA, 2006. DEMO

• Karel Zimmermann, Tomas Svoboda. Probabilistic Estimation of Articulated Body Model from Multiview Data. In Proceedings of 3rd European Medical and Biological Engineering Conference. International Federation for Medical and Biological Engineering. ISSN 1727-1983, volume 11, November 2005. [PDF].

• Tomas Svoboda, Daniel Martinec, and Tomas Pajdla. A convenient multi-camera self-calibration for virtual environments. PRESENCE: Teleoperators and Virtual Environments, pp 407-422, 14(4), August 2005. MIT Press Manuscript of the journal article [ PDF, 1.7MB ]. Its bib-entry.
• J.Matas, K.Zimmermann, A New Class of Learnable Detectors, In Proceedings of Scandinavian Conference on Image Analysis (SCIA'05), pp. 541-550, 2005

• T. Svoboda. A Software for Complete Calibration of Multicamera Systems. In Image and Video Communications and Processing, Proceedings od SPIE--IS&T Electronic Imaging, 2005, SPIE vol. 5685, pp. 115-128

Talks

• Dagstuhl seminar. Human Motion - Understanding, Modeling, Capture and Animation. 13th Workshop "Theoretical Foundations of Computer Vision". Talk [PDF]. Note that hyperlinked videos are on-line.