/face_rec_MobileFaceNet

Depthwise Convolution to SphereFace

Primary LanguageJupyter Notebook

MobileFaceNet

Contributions:

  1. Add Depthwise Convolution to SphereFace
  2. We employed the MobileFaceNet architecture and trained using the angular softmax (A-Softmax) loss function on the VGGFace2 dataset.

Model size: 8MB. Feature size: 256-dimension. Accuracy: 99.30% on LFW.

Citation

If you use this code for your research, please cite our papers.

@inproceedings{SGGAN2019,
  title={Matching Thermal to Visible Face Images Using a Semantic-Guided Generative Adversarial Network},
  author={Chen, Cunjian and Ross, Arun},
  booktitle={IEEE International Conference on Automatic Face & Gesture Recognition},
  year={2019}
}

References:

1. https://github.com/wy1iu/sphereface
2. http://www.robots.ox.ac.uk/~vgg/data/vgg_face2/