GPU aXelerated Face Identification
To make: Go to /code/src directory,
mkdir obj
make
To save eigenfaces and reconstructed faces to disk, you have to create the 'eigen' and 'reconstructed' directories beforehand (in /code/src)
To run the GPU version:
./FaceIdX
To run the CPU version:
./FaceIdX -cpu
The paths can be relative. E.g., if FaceIdX is run from /code/src, you can type ../../data/demo/training to select the training set
Using 2 single-file public domain librairies for image reading/writing: https://github.com/nothings/stb.git by Sean T. Barrett
The faces in the dataset are from:
- The Yale dataset of faces
- Internet (for celebraties)
- Our pictures (for us)
Francis Lan and XiangLu Kong
Project for W4995 - GPU Computing, Spring 2015