This repository contains the implementation of the metrics used in the paper:
A. Elobaid, N. Ramoly, L. Younes, S. Papadopoulos, E. Ntoutsi, and I. Kompatsiaris, "Sum of Group Error Differences: A Critical Examination of Bias Evaluation in Biometric Verification and a Dual-Metric Measure," 2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG), Istanbul, Turkiye, 2024, pp. 1-9, doi: 10.1109/FG59268.2024.10582012.
To promote accessibility and encourage exploration of fairness in biometric verification, we have hosted the implementation of the metrics on a Hugging Face Space. This space provides an easy-to-use interface for researchers, practitioners, and the broader biometric fairness community to utilize the metrics described in the paper.
The metrics aim to evaluate bias comprehensively by introducing the Sum of Group Error Differences (SGED), a dual-metric measure that critically examines biases in biometric systems.
The implementation is available here:
Hugging Face Space: Bias Evaluation in Biometric Verification
- Accepts distance and label data in the form of a pickle file.
- Outputs the computed metrics for bias evaluation.
- Easy-to-use interface for evaluating bias metrics.
- Explanation and examples of the metrics used.
- Open for the biometric fairness community to use and provide feedback.
For detailed explanations of the metrics and how they are calculated, visit the space's README file.
This documentation provides an in-depth understanding of the evaluation criteria and their importance in ensuring fairness in biometric systems.
If you use the implementation or metrics in your work, please cite the following paper:
@inproceedings{elobaid2024sum,
author={A. Elobaid and N. Ramoly and L. Younes and S. Papadopoulos and E. Ntoutsi and I. Kompatsiaris},
title={Sum of Group Error Differences: A Critical Examination of Bias Evaluation in Biometric Verification and a Dual-Metric Measure},
booktitle={2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG)},
year={2024},
pages={1-9},
doi={10.1109/FG59268.2024.10582012}
}