insightface

The training process of Subcenter ArcFace is almost same as ArcFace, except for one extra hyperparameter (subcenter number loss_K) to relax the intra-class compactness constraint. In our experiments, we find loss_K=3 can achieve a good balance between accuracy and robustness.

2. Training Dataset

MS1MV0 (The noise rate is around 50%), download link (baidulink, code 8ql0) (googledrive)

3. Training Steps

1). Train Sub-center ArcFace (loss_K=3) on MS1MV0.

2). Drop non-dominant subcenters and high-confident noisy data (>75 degrees).

python drop.py --data <ms1mv0-path> --model <step-1-pretrained-model> --threshold 75 --k 3 --output <ms1mv0-drop75-path>

3). Train ArcFace on the new MS1MV0-Drop75 dataset.

Citation

If you find Sub-center ArcFace useful in your research, please consider to cite the following related papers:

@inproceedings{deng2020subcenter,
  title={Sub-center ArcFace: Boosting Face Recognition by Large-scale Noisy Web Faces},
  author={Deng, Jiankang and Guo, Jia and Liu, Tongliang and Gong, Mingming and Zafeiriou, Stefanos},
  booktitle={Proceedings of the IEEE Conference on European Conference on Computer Vision},
  year={2020}
}