Skip to content

FEI Morph Dataset

Description

FEI Morph Dataset is a dataset generated using the images contained in the FEI Face Database.

We suggest using this dataset for the evaluation of new Differential Morphing Attack Detectors (D-MAD), using the couples that can be downloaded here (with both the criminal and the accomplice).

About the pair selection, for each subject, five other subjects of the same ethnicity and gender have been selected for morphing; in order to maximize the attack potential of the morphed images (i.e. the probability of fooling FRSs), the average face verification scores of three commercial SDKs Verilook, Cognitec, and Innovatrics have been used to select the most similar subjects for each individual. 

Main Features:
  • 200 subjects, equally split between male and female.
  • All faces are mainly represented by subjects between 19 and 40 years old.
  • Variety in appearances, hairstyles, and accessories.

Version V1

This dataset has been obtained with three different morphing algorithms:

  • [C02] FaceFusion [1]
  • [C03] FaceMorph proposed by the Norwegian University of Science and Technology in [1]
  • [C08] a variant of the triangulation with STASM-landmarks proposed by the University of Twente in [1]

[1] G. Borghi, et al. Revelio: A modular and effective framework for reproducible training and evaluation of morphing attack detectors. IEEE Access


Version V2 (extended version with additional morphing algorithms w.r.t. Version V1)

In addition to the morphing algorithms in V1, this dataset includes four additional morphing algorithms:

  • [C01] FaceMorpher [2]
  • [C05] UBOMorph proposed by the University of Bologna in [1], available at this link.
  • [C15] Developed by SURYS
  • [C16] proposed by the University of Twente in [3]

[2] K. Raja, et al. Morphing attack detection-database, evaluation platform, and benchmarking. IEEE TIFS
[3] I. Batskos, et al. Visualizing landmark-based face morphing traces on digital images. Frontiers in Computer Science


Version Temporal (V3, extended version with additional morphing algorithms w.r.t. Version V2)

In addition to the morphing algorithms in V1, this dataset includes four additional morphing algorithms:

  • [C21] inspired by [4]
  • [C22] inspired by [5]
  • [C24] Arc2Morph [6]
  • [C25] proposed by the University of Twente in [7]
  • [C26] proposed by Hochschole Darmstadt and NTNU in [8]

[4] M. Grimmer, et al. Ladimo: Face morph generation through biometric template inversion with latent diffusion. IEEE IJCB
[5] R. Ismayilov, et al. StyleDemorpher: high-quality face demorphing via StyleGAN2’s latent space. Machine Vision and Applications
[6] N. Di Domenico, et al. Arc2Morph: Identity-Preserving Facial Morphing with Arc2Face. IEEE FG
[7] I. Batskos, et al. Improving fully automated landmark-based face morphing. IWBF
[8] W. Kabbani, et al. StableMorph: High-Quality Face Morph Generation with Stable Diffusion. IEEE IJCB


Release of the versions (V1, V2 and Temporal)

  • Download and sign the agreement (single for the three versions).
  • Send the agreement to Guido Borghi and/or Annalisa Franco and/or Matteo Ferrara
  • Here, you can find the pairs we used for the D-MAD task
  • Naming file convention: M_<subj2>_<subj2>_<morphing_algorithm>_<morph_factor>_<morph_factor>_<post_proc_auto>_<post_proc_manual>_<digital/P&S>

We thank the authors of the original FEI Face database, and in particular Dr. Carlos Eduardo Thomaz.

References

If you use these datasets, please cite the following papers:

@inproceedings{di2023combining,
  title={Combining Identity Features and Artifact Analysis for Differential Morphing Attack Detection},
  author={Di Domenico, Nicol{\`o} and Borghi, Guido and Franco, Annalisa and Maltoni, Davide},
  booktitle={International Conference on Image Analysis and Processing},
  pages={100--111},
  year={2023},
  organization={Springer}
}

@inproceedings{pellegrini2025foundation,
  title={Foundation Models across Time: Benchmarking Detection of Evolving Face Morphing Attacks},
  author={Pellegrini, Lorenzo and Pandolfini, Serafino and Di Domenico, Nicol{\`o} and Borghi, Guido and Franco, Annalisa and Ferrara, Matteo and Maltoni, Davide},
  booktitle={ECCV 2026 Workshop on Foundation and Generative Models in Biometrics},
  year={2025}
}
@article{thomaz2010new,
  title={A new ranking method for principal components analysis and its application to face image analysis},
  author={Thomaz, Carlos Eduardo and Giraldi, Gilson Antonio},
  journal={Image and vision computing},
  volume={28},
  number={6},
  pages={902--913},
  year={2010},
  publisher={Elsevier}
}