A hand pose estimator using model fitting by finding minimal energy, using OpenCV and tnc.c library.
C C++
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
cmake/Modules
CMakeLists.txt
README
bg_fg_blobs.cpp
bg_fg_blobs.h
freenect_ocv.cpp
hand_data.h
main.cpp
output.avi
simple_tnc.c
tnc.c
tnc.h

README

OpenHPE is an open source Hand Pose Estimator using OpenCV, tnc.c, libfreenect.
It is developed in the Fluid Interfaces group of the MIT Media Lab.

initial results:
1. http://www.youtube.com/watch?v=uETHJQhK144
2. http://www.youtube.com/watch?v=PO1KPdiG2us

(Vauge) Build instructions
- Create a project in your favorite DevEnv (MSVS2008/2010, XCode, CodeBlocks, Eclipse, a Makefile ?!)
- Add all these .c,.cpp,.h files into the project
- Add OpenCV and libfreenect include directories and library directories
- Build & Run

Concrete Build instructions
- run `cmake .`, make sure CMake knows which build system you want (Unix Makefile, XCode, MSVS, etc), and OpenCV libraries are well defined
- run `make` / build in your IDE

This is not for complete beginners, but a beginners will benefit a lot from trying to build and tinker around. Please do so.

Walkthrough of the code:
http://www.morethantechnical.com/2010/12/28/hand-gesture-recognition-via-model-fitting-in-energy-minimization-wopencv/