HyperFace

Description

We introduce HyperFace, a novel framework for generaing synthetic face recognition datasets and increasing the inter-class variation. We formulate the dataset generation as a packing problem on the embedding space (represented on a hypersphere) of a face recognition model and propose a new synthetic dataset generation approach (called HyperFace). We formalize our packing problem as an optimization problem and solve it with a gradient descent-based approach. Then, we use a conditional face generator model to synthesize face images from the optimized embeddings. We release several synthetic datasets of face images up to 50,000 unique synthetic identities and 3.2 million images.

Project page: https://www.idiap.ch/paper/hyperface

Reference

If you're using this dataset, please cite the following publication:

@inproceedings{shahreza2025hyperface,
          title = {HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere},
         author = {Hatef Otroshi Shahreza and S{\'e}bastien Marcel},
      booktitle = {The Thirteenth International Conference on Learning Representations},
           year = {2025}
}