VOSM (Vision Open Statistical Models, C++)

 

Download

     
– vosm-0.3.2.tar.bz2 (Linux); vosm-0.3.2.zip (Windows)
Current VOSM implements all popular ASMs/AAMs, and one new fitting methods (namely, 2D profile ASM) both for static images and dynamic image sequences, including:
  1. 1D profile ASM
  2. 2D profile ASM
  3. Direct Local texture contrained (LTC) ASM
  4. Basic AAM
  5. ICIA(Inverse compositional image alignment) AAM
  6. IAIA(Inverse additive image alignment) AAM

The only dependent libraries of vosm-0.3.2 are

  1. OpenCV Subversion 6267 or above
  2. boost-1.42 or above (specifically, boost-regex and boost-filesystem) or above

VOSM 0.3.2 is a stable version, which has been tested with Boost 1.42 and OpenCV SVN 6267, both in Windows 7 and Ubuntu 11.04.

 

VOSM Usage

  1. First of all, in order to use VOSM, you need to find and download respective ShapeInfo.txt files summarized in the table at http://www.visionopen.com/resources/computer-vision/statistical-models/ .
  2. How to use VOSM to train a face model and fit an untrained face using the trained face model? Learn to use two commands smbuilding and smfitting. Tow examples are given as:
    1. smbuilding 
      -a “/media/WD2T/Research/Databases/face/2D/EMOUNT/annotations/training” 
      -i “/media/WD2T/Research/Databases/face/2D/EMOUNT/images/training” 
      -s “/media/WD2T/Research/Databases/face/2D/EMOUNT/annotations/training/ShapeInfo.txt” 
      -d “EMOUNT” 
      -c 1 
      -t “SMNDProfile”
    2. smfitting 
      -o “/media/WD2T/Research/Databases/face/2D/JIAPEI/traineddata/33Frontal/BGR” 
      -t “ASM_PROFILEND” 
      -i “/media/WD2T/Research/Databases/face/2D/JIAPEI/images/testing” 
      -a “/media/WD2T/Research/Databases/face/2D/JIAPEI/annotations/testing” 
      -d “JIAPEI” 
      -s “true” 
      -r “true”

More details about these two commands can be found by respective help commands.

 

AAM Building

– AAMBuildingGUI-Toronto.rar
Active Appearance Model (AAM) Building GUI binary code.
A wxWidget based GUI tool is afforded to simplify the process of building an AAM model. Since this tool was roughly finished in Toronto, Canada, I call it a Toronto version. 
– aambuilding-1.0.zip
Active Appearance Model (AAM) Building Source Code.
You can obtain some basic AAM concepts from here.
aambuilding-1.0 is a stable version of AAM Building, which has been tested with Boost 1.36 and OpenCV-cvs, both in Windows Vista and Ubuntu 8.04. 

Latest Forum Discussions