Face Recognition using a CNN architecture.
Project uses Yale dataset as a demo, but is extensible to any face recognition dataset. Download Yale Face Database from here.
- Do
pip install -r requirements.txt
to install all deps. - Get the database as mentioned above. Run
python utils.py <yale_dataset_folder> <data_folder>
to create train and val data fromyale_dataset_folder
intodata_folder
. - Run
giftopng.sh <folder>
to convert all the images inside the subfolders of<folder>
into pngs. - Run using
python main.py --data <data_folder>
wheredata_folder
contains folderstrain
andval
containing training and validation data respectively. - Inside
train
andval
folders, program expects folder for each of the labels and these folder contains samples for these labels.
Add picture for the architecture and tune it.
MIT