Note
Cite the original paper if you use the implementation on this page!
Important
It is forbidden to use the following S-MAD and D-MAD models on FVC and NIST benchmarks without the consent of the authors.
-
A double siamese framework for differential morphing attack detection (Borghi et al., Sensors 2021)
- Original paper
- Resource
- Code: Python, PyTorch
- Training datasets: PMDB, MorphDB, Idiap-morph datasets
- Notes: official implementation, tested on FVC (SOTAMD_D-1.0 EER = 23.37%)
-
Combining identity features and artifact analysis for Differential Morphing Attack Detection (Di Domenico et al., ICIAP 2023)
- Original paper
- Resource
- Training datasets: PMDB, MorphDB, Idiap-morph
- Notes: official implementation, tested on FVC (SOTAMD_D-1.0 EER = 10.23%)
-
Dealing with Subject Similarity in Differential Morphing Attack Detection (Di Domenico et al., under submission)
- The paper is currently under submission
- Resource
- Training datasets: PMDB, MorphDB, Idiap-morph
- Notes: official implementation, tested on FVC (SOTAMD_D-1.0 EER = 7.84%)
-
Deep Face Representations for Differential Morphing Attack Detection (Scherag et al., TIFS 2020)
- Original paper
- Resource
- Code: Python, PyTorch, MXNet
- Training datasets: PMDB
- Notes: unofficial implementation, tested on FVC (MORPHDB_D-1.0 EER = 0.0%)
- UBO-SMAD-R3: an Inception-ResNet-based model for Single-image Morphing Attack Detection
- Resource
- Code: Python, PyTorch
- Training datasets: PMDB, MorphDB, Idiap-morph, Chimo
- Notes: tested on FVC-onGoing (SOTAMD_D-1.0 EER = 10.33%)
- Revelio framework
- Revelio is a framework to simplify D-MAD and S-MAD model development
- Resource
- Documentation
- FVC-onGoing platform
- Link
- S-MAD and D-MAD evaluations on sequestered datasets.
- FRVT MORPH
- Link
- S-MAD and D-MAD evaluations on sequestered datasets.
- Public
- FEI Morph Dataset A dataset for the S/D-MAD task.
- ChiMo Dataset A dataset for the S-MAD task.
- Idiap-morph
- Private
- PMDB
- MorphDB (only for evaluation on the FVC platform)