Face-Recognition icon indicating copy to clipboard operation
Face-Recognition copied to clipboard

Face Recognition project in PyTorch using CNNs

Face Recognition

Face Recognition using a CNN architecture.

Dataset

Project uses Yale dataset as a demo, but is extensible to any face recognition dataset. Download Yale Face Database from here.

Running

  • 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 from yale_dataset_folder into data_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> where data_folder contains folders train and val containing training and validation data respectively.
  • Inside train and val folders, program expects folder for each of the labels and these folder contains samples for these labels.

TODO

Add picture for the architecture and tune it.

License

MIT