insightface/recognition/idmmd

Physical-based Rendering for NIR-VIS Face Recognition

by Yunqi Miao*, Alexandros Lattas*, Jiankang Deng, Jungong Han, and Stefanos Zafeiriou.

For more information, please check our

[Arxiv] [NeurIPS 2022 Paper]

If you find this project useful in your research, please consider citing:

@inproceedings{miao2022physically,
  title={Physically-Based Face Rendering for NIR-VIS Face Recognition},
  author={Miao, Yunqi and Lattas, Alexandros and Deng, Jiankang and Han, Jungong and Zafeiriou, Stefanos},
  booktitle={NeurIPS 2022},
  year={2022}
}

Overview

Training

For this project, we used python 3.7.10.

How to run?

sh run.sh

Testing

Preparation

• Downloading data (112 x 112) from [Google drive]
  • Put data to data/$dataset_name
  • Copy data list from here

Note that: casia(fold_1) is provided for research purposes only. For the rest data, please refer to the original publications.

• Downloading models from [Google drive]
  • Put pretrain model at models/pretrain/
  • Put finetune model at models/finetune/$dataset/

How to run?

sh eval.sh