Skip to content

Latest commit

 

History

History
94 lines (64 loc) · 5.19 KB

README.md

File metadata and controls

94 lines (64 loc) · 5.19 KB
Raw

VOSM

What does VOSM stand for?

Vision Open Statistical Model (VOSM) contains variants of 2D Statistical Models, namely, variants of ASMs and AAMs. It's mainly composed of two parts:

  • Building: build the statistical models (both 2D and 3D are supported in VOSM). For now, the following models can be built by specifying different parameters:

    • Shape Model: -t "SM", fundamental shape model
    • Texture Model: -t "TM", fundamental texture model
    • Appearance Model: -t "AM", fundamental appearance model, originated from Basic AAM
    • Image Alignment appearance Model: -t "IA", inverse compositional image alignment, for ICIA AAM
    • Feature Model: -t "FM", a generalized model based on ANY type of features adopted by the users
    • Shape model with Local Texture Constraints: -t "SMLTC", originated from CLM
    • SMNDPROFILE. -t "SMNDPROFILE", including original 1D profile ASM and 2D profile ASM proposed by me, please check chapter 3 of my PhD thesis.

  • Fitting: test the effects of fitting (ONLY 2D fitting is suppprted in VOSM)

Current Version

0.3.5 is our FIRST release.

Annotated Datasets

Datasets No. Of Annotated Points
IMM 58
AGING 68
BIOID 68
FRANCK/Talking Face 68
XM2VTS 68
UMDFaces 21

Prerequisites

Download, Compile, and Build

$ git clone https://github.com/jiapei100/VOSM
$ mkdir build
$ cd build
$ ccmake ../
$ make -j8
$ sudo make install

Two commands and eight libraries will be built:

By default, 2 commands will be installed under: /usr/local/bin:

  • testsmbuilding
  • testsmfitting

By default, 8 libraries will be installed under: /usr/local/lib:

  • libvosm_comalgs.so
  • libvosm_common.so
  • libvosm_cvcommon.so
  • libvosm_ensembletraining.so
  • libvosm_featureextraction.so
  • libvosm_integraltransform.so
  • libvosm_smbuilding.so
  • libvosm_smfitting.so

How to use VOSM?

Please refer to our wiki How to use VOSM?.

Cascade Files for Face Components

In file testsmfitting, 4 cascade files are used to detect 4 different face parts:

which can be found in opencv_contrib.

Key Relative Publications

  1. P. JIA, 2D Statistical Models, Technical Report of Vision Open Working Group, 2st Edition, Oct 21, 2010.

  2. P. JIA. Audio-visual based HMI for an Intelligent Wheelchair. PhD thesis, University of Essex, 2010.

  3. T. Cootes and C. Taylor. Statistical models of appearance for computer vision. Technical report, Imaging Science and Biomedical Engineering, University of Manchester, March 8 2004.

  4. I. Matthews and S. Baker. Active appearance models revisited. International Journal of Computer Vision, 60(2):135–164, November 2004.

  5. M. B. Stegmann, Active Appearance Models: Theory, Extensions and Cases, 2000.