Usage

How to use facepointChooser.py, faceMorpher, and stochopt.py (simulated annealing) Example commands Files (results/ testing)

** facepointChooser.py

$ python facepointChooser.py -l -i [ …]

All chosen coordinates will be logged in one log file. Simply copy and paste each relevant section (headed by “ **”) into separate log files, and put the files into the same directory as the faceMorpher build. Sample log files are included for reference/ testing use.

** faceMorpher

For basic help information, just run $ ./faceMorpher for help information. Note: options can be input in any order. The executable was built in Mac 10.7+, OpenCV 2.4.5. Put executable in the same directory as images/ landmarks files (if morphing), and same directory as folders (if getting mean/ eigenfaces)

Possible compilation issues (XCode settings): . Header Search Paths: /usr/local/include . Library Search Paths: /usr/local/lib . C++ Standard Libary: libstdc++ (GNU C++ standard library)

Prepared files for results:

Morphing (goto test0 dir): . 01happy.jpg, 01ref.txt, 02centerlight.jpg, 02ref.txt . results in test0_movie

Used to make rmMorphFast.mp4 (goto rm dir): . Images and landmark files in rm directory. . results in rm_movie

Mean face/ Eigenfaces (goto mean_eigen dir): . images in cleanMeanData directory, landmarks in meanFiles directory . results generated can be organized with script "mrClean". $ bash mrClean

• Facemorphing $ ./faceMorpher -i0 -f0 <image0’s landmarks> -i1 -f1 <image1’s landmarks>

EXAMPLE: ./faceMorpher -i0 01happy.jpg -f0 01ref.txt -i1 02centerlight.jpg -f1 02ref.txt

• Mean face $ ./faceMorpher -m .// -mf ./<faces-for-mean’s landmarks directory>/

EXAMPLE: ./faceMorpher -m ./cleanMeanData/ -mf ./meanFiles/ -e 1

• Eigenfaces $ ./faceMorpher -m .// -mf ./<faces-for-mean’s landmarks directory>/ -e <1|0>

-e 1: eigenface with normalized shape (for AAMs) -e 0: eigenface directly from images (normal)

**Simulated annealing

$ python stochopt.py

Dependencies: Python 2.7+, Tkinter