The Face Annotation Interface - Java toolkit for facial recognition using eigenfaces.
The Faint application was originally written by Malte Mathiszig as a Java application. Initially developed in the context of a Bachelor Thesis at the University of Oldenburg, faint has been integrated into several projects maintained by the OFFIS Institute for Information Technology. To attract a broader audience, the source code has been released under GNU General Public License (GPL) in October 2007 and hosted at http://faint.sourceforge.net/.
The original sourceforge release was feature-rich. It included a photo browser EXIF annotations, Image caches and a host of other features. These features were powerful and made faint a valuable library. At the same time however, they obscured the most valuable feature, its Eigenfaces implementation.
OpenCV face detection was written around the 1.0 version of the Windows DLL and had not been ported. The Betafaces implementation seems to be no longer active. Ultimately this left it unusable on platforms other than windows.
This fork of Faint attempts to simplify and reduce the library to its core. I have left in the face database support and thumbnail caching as they make the program useful by itself. Ultimately though it would be preferred to allow library users to implement their own storage mechanism.
This version of the library includes a JNI wrapper for OpenCV 2.0 and is targeted for Mac OSX. It expects a C++ build environment.
You should build the OpenCV toolkit from source. Information on this can be found at http://opencv.willowgarage.com/wiki/Mac_OS_X_OpenCV_Port. This also contains the data for the default cascades.
Next get the source for this library
git clone git://github.com/jeffrafter/faint.git
Once you have it, go ahead and build:
cd faint ./build
This should build everything including the JNI wrapper.
Now that you have it installed you can run the samples:
cd samples ./detect ./train ./recognize
The detect application will utilize the OpenCV wrapper to execute the detection pattern based on the included cascade file in the samples folder. This basically tries to find things that match a large training set of objects (which happen to be faces). It will print out the regions (x, y, width, height) it finds in the image.
You can use these to start training the system. Run the train application, pass it an image file, the region information and an annotation (basically the person's name from the region). This adds the region to the thumbnail cache and database and marks it as an image that should be used for training the eigenfaces algorithm.
Finally, you can submit an image and region to the recognize application and see if it matches any of the training annotations you entered. The results will print out with the annotation names and their weights. The one with the highest weight generally wins.
/*******************************************************************************
-
- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- +
- | |
- faint - The Face Annotation Interface
- | Copyright (C) 2007 Malte Mathiszig |
- | This program is free software: you can redistribute it and/or modify |
- it under the terms of the GNU General Public License as published by
- | the Free Software Foundation, either version 3 of the License, or |
- (at your option) any later version.
- | |
- This program is distributed in the hope that it will be useful,
- | but WITHOUT ANY WARRANTY; without even the implied warranty of |
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- | GNU General Public License for more details. |
- | You should have received a copy of the GNU General Public License |
- along with this program. If not, see http://www.gnu.org/licenses/.
- | |
-
- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- + *******************************************************************************/
Special thanks to Malte Mathiszig for releasing an amazing application.