Presentation Attack Detection on Chile realistic ID Card dataset
0.85EERSigLIP-SO400M (Unimodal)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SigLIP-SO400M (Unimodal)Modality=Unimodal, Model Size=SO400M2026.06 | 0.85 | 0 | 0 | 0.43 | |
| SmolVLM2-2.2B-Gen.Setting=Generative, Model Size=2.2B2026.06 | 0.93 | 0.34 | 0.34 | 0.76 | |
| SmolVLM2-500M-Gen.Setting=Generative, Model Size=500M2026.06 | 1.96 | 0.17 | 0.68 | 6.23 | |
| DenseNet2026.06 | 2.21 | 0.51 | 0.68 | 7.77 | |
| SmolVLM2-2.2B-Disc.Setting=Discriminative, Model Size=2.2B2026.06 | 17.08 | 30.03 | 45.69 | 77.71 | |
| SmolVLM2-500M-Disc.Setting=Discriminative, Model Size=500M2026.06 | 25.02 | 56.7 | 74.85 | 91.63 | |
| SmolVLM2-2.2B-Zero-ShotSetting=Zero-shot, Model Size=2.2B2026.06 | 45.26 | 85.22 | 91.12 | 97.36 | |
| SmolVLM2-500M-Zero-ShotSetting=Zero-shot, Model Size=500M2026.06 | 50.46 | 90.78 | 95.3 | 99.15 |