GestureDetector trains a gesture SVM classifier by extracting HOG descriptors from input images. Given a model, it can also determine which gesture it is in a given image.

To build and run this project, you should prepare:

• OS X 10.9 or above
• Xcode 5.0 or above
• OpenCV 2.4.2 or above
• At least 1GB memory

This project uses Sebastien Marcel - Hand Posture and Gesture Datasets for gesture samples, and background images from INRIA Person Dataset for negative samples. You can download the dataset here.

The classifier used in this project is SVM; particularly, LIBSVM. The three binary files in "libsvm" directory were compiled using LIBSVM source code under OS X.

1. Start a new XCode Command Line Tool project.
2. Rght click on the project, and click "Add files to...".
3. When Finder pops up, hit "/" to bring up the navigation panel.
4. Type in "/usr/local/lib".
5. Add in all the .dylib files that you need. To prevent linker errors, I recommend you initially add ALL the files ending in "...2.4.0.dylib". Now, you should have a bunch of .dylib files in your project. Feel free to move them to a separate group within your project.
6. Click on the project file and go to "Build Settings".
7. Search for "Header Search Paths", and change the path to "/usr/local/include".
8. Search for "Library Search Paths", and change the path to "/usr/local/lib".
9. [For Xcode 5] Search for "C++ Standard Library", and change to "libstdc++ (GNU C++ standard library".