Cross-Quality Labeled Faces in the Wild (XQLFW)
Here, we release the database, evaluation protocol and code for the following paper:
đź“‚
Database and Evaluation Protocol
If you are interested in our Database and Evaluation Protocol please visit our website.
đź’»
Code
We provide the code to calculate the accuracy for face recognition models on the XQLFW evaluation protocol.
🥣
Requirements
🚀
How to use
- Download the database and evaluation protocol here
- Inference the images and save the embeddings and labels to a numpy file (*.npy) according to:
[[pair1_img1_embed, pair1_img2_embed, pair2_img1_embed, pair2_img2_embed, ...], [True, True, False, ...]]
- Run the evaluate.py code with
--source_embedding
argument containing the absolute path to a directory containing your embedding .npy files:python evaluate.py --source_embeddings="path/to/your/folder" --csv --save
- Use the flag
--csv
if you want to get the results displayed in csv instead of a table. - Use the flag
--save
to save the results into the source_embedding directory.
- Use the flag
- See the results and enjoy!
đź“–
Cite
If you use our code please consider citing:
@misc{knoche2021crossquality,
title={Cross-Quality LFW: A Database for Analyzing
Cross-Resolution Image Face Recognition in Unconstrained Environments},
author={Martin Knoche and Stefan Hörmann and Gerhard Rigoll},
year={2021},
eprint={2108.10290},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
and mabybe also:
@TechReport{LFWTech,
author={Gary B. Huang and Manu Ramesh and Tamara Berg
and Erik Learned-Miller},
title={Labeled Faces in the Wild: A Database for Studying
Face Recognition in Unconstrained Environments},
institution={University of Massachusetts, Amherst},
year={2007},
number={07-49},
month={October}
}
@TechReport{LFWTechUpdate,
author={Huang, Gary B and Learned-Miller, Erik},
title={Labeled Faces in the Wild: Updates and New
Reporting Procedures},
institution={University of Massachusetts, Amherst},
year={2014},
number={UM-CS-2014-003},
month={May}
}
✉️
Contact
For any inquiries, please open an issue on GitHub or send an E-Mail to: [email protected]