Introduction


What is it ?
  • The MPT stands for "Machine Perception Toolbox" and is pronounced ``Empty'', ``Empty Box'', or "Mu". It supplies cross-platform libraries for real-time perception primitives, including face detection, eye detection, blink detection, color tracking. Soon it will also include expression recognition, predictive color tracking, and tracking based on multisensor fusion. In addition, it supplies many example which show how to embed the MPT functions in applications. It also provides applications (eyedetection, real time face tracking in video ...) which are useful applications in their own right. Finally, Matlab .mex interfaces are provided for many of these core libraries for those who like using Matlab. See Also :
    • The Screenshots section shows examples of applications where MPT can be typically used.
    • The Download section is the place to get the latest version of MPT package.
    • The Manual (in pdf) provides guidelines on how to install MPT, and examples of how to use MPT.
    Authors :
  • Ian Fasel
  • Ryan Dahl
  • John Hershey
  • Bret Fortenberry
  • Josh Susskind
  • Javier R. Movellan
    • License :
    The MPT is distributed under The BSD License (Free Software) This license is compatible with the Free Software Foundation's GPL license.
    Quick History :
  • In May 2001 John Hershey, a member of UCSD's Machine Perception Laboratory (MPLab), wrote "viola.m", a MatLab function that performed image filtering based on the integral image trick that had been recently rediscovered by Viola and Jones.
  • In August 2001 Ian Fasel, another member of the MPLab rewrote part of the code in Matlab resulting on "violaSearch.m"
  • In Apr 2002 Ryan Dahl, with funding from the MPLab and guidance from Ian Fasel wrote "viola++", which became the core for mpisearch.
  • In Feb 2003, Ian Fasel wrote the first version of "mpisearch" based on "viola++".
  • Several versions of mpisearch evolved in parallel as they were applied for different problems (color tracking, face tracking, eye detection) and under different platforms (windows, Linux, Mac OS X).
  • During this time the main developer was Ian Fasel with some contributions from Josh Susskind and Brett Fortenberry, which were members of the MPLab.
  • In the Fall 2004 we put effort into unifying the different versions of mpisearch, collecting the different applications that had developed over the years, develop example code and documentation and bringing it all together under as a package that we decided to name "MPT".
  • The MPT was originally developed at UCSD's Machine Perception Laboratory (MPLab) under the direction of Javier R. Movellan. Javier's role was to advise on the applications of the research and to focus the application of the software within the framework of probabilistic inference and stochastic processes. He also worked on the documentation of the toolbox and helping it come together as a unified product.
    • Copyrights :
    Copyrights (C) 2000->2005, Machine Perception Laboratory, University of California San Diego.