/octuplet-loss

Repo for our Paper: Octuplet Loss: Make Your Face Recognition Model Robust to Image Resolution

Primary LanguagePythonMIT LicenseMIT

Octuplet Loss - Make Face Recognition Robust Against Image Resolution

Code style: black License Last Commit

Here, we release our code and model utilized in the following paper:

Loss Visualization

🏆 Performance (Accuracy [%])

All models are finetuned with Octuplet-Loss:

Model LFW XQLFW
ArcFace 99.55 93.27
MagFace 99.63 92.92
FaceTransformer 99.73 95.12

💻 Code

We provide the code of our proposed octuplet loss:

🧠 Model

We provide the model via huggingface https://huggingface.co/Martlgap/FaceTransformerOctupletLoss

🥣 Requirements

Python 3.8

📖 Cite

If you use our code please consider citing:

@inproceedings{knoche2023octuplet,
  title={Octuplet loss: Make face recognition robust to image resolution},
  author={Knoche, Martin and Elkadeem, Mohamed and H{\"o}rmann, Stefan and Rigoll, Gerhard},
  booktitle={2023 IEEE 17th International Conference on Automatic Face and Gesture Recognition (FG)},
  pages={1--8},
  year={2023},
  organization={IEEE}
}

✉️ Contact

For any inquiries, please open an issue on GitHub or send an E-Mail to: Martin.Knoche@tum.de