Multi-IVE: Privacy Enhancement of Multiple Soft-Biometrics in Face Embeddings

This repository contains the source code to reproduce the following paper:

@inproceedings{multi-ive,
  title={Multi-IVE: Privacy Enhancement of Multiple Soft-Biometrics in Face Embeddings},
  author={Melzi, Pietro and Shahreza, Hatef Otroshi and Rathgeb, Christian and Tolosana, Ruben and Vera-Rodriguez, Ruben and Fierrez, Julian and Marcel, S{\'e}bastien and Busch, Christoph},
  booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
  pages={323--331},
  year={2023}
}

Usage

To create a virtual environment run the command:

$ conda env create --name envname --file=environment.yml

Training

To generate masks that represent the order of feature elimination run the command:

python main.py

Masks can be generated for the following settings:

• no domain transformation
• pca domain transformation with k={0, 3, 5}
• ica domain transformation

Evaluation

Use the file evaluate_IVE.py to evaluate Multi-IVE as in the paper.

Use the file first_components_pca.py to evaluate Multi-IVE in the scenario of principal component elimination in the order of their importance.

Use the file rdm_mask.py to evaluate Multi-IVE in the scenario of random feature elimination.