Published November 11, 2024 | Version v1
Dataset Open

Digi2Real

  • 1. ROR icon Idiap Research Institute

Description

Description

Digi2Real is a synthetic face dataset containing images of 20,000 unique synthetic identities. We introduce a novel framework for realism transfer, designed to enhance the realism of synthetically generated face images. This framework is applied to a subset of the DigiFace 1M dataset to produce photorealistic images, which are more effective for training face recognition models than the original DigiFace dataset. By combining the controllable features of the graphics pipeline with our realism enhancement technique, we present a new approach for creating attribute-controllable face recognition datasets.

 

License

Digi2Real dataset is published for non-commercial research use only, please check the license agreement.  

 

Reference

If you use this dataset, please cite the following publication:

 @article{george2024digi2real,
          title={Digi2Real: Bridging the Realism Gap in Synthetic Data Face Recognition via Foundation Models}, 
          author={Anjith George and Sebastien Marcel},
          year={2024},
          eprint={2411.02188},
          url={https://arxiv.org/abs/2411.02188}, } 

Files

Digi2Real_Documentation.pdf

Files (7.7 GB)

Name Size Download all
md5:0586c4c25048eaf24ebc883c45b73fe1
2.9 GB Download
md5:6416df0ce019697c56c0bf108afa63ba
540.2 kB Preview Download
md5:f1ccb2f69597b45bf4ee48fa38682170
4.8 GB Download
md5:1de5933fe456a10e9c747ee829cfa272
3.8 kB Preview Download

Additional details

Related works

Is described by
Journal article: 10.48550/arXiv.2411.02188 (DOI)

Funding

Hasler Foundation
reSponsible fAir FacE Recognition (SAFER) 21044