Existing datasets
Model Metrics
-- | predy_0 | predy_1 |
gt_0 | TN | FP |
gt_1 | FN | TP |
- Attack Presentation Classification Error Rate[APCER] : FP / (TN + FP)
- Bona Fide Presentation Classification Error Rate[BPCER] : FN / (TP + FN)
- Average Classification Error Rate[ACER] : (APCER + BPCER) / 2.0
- False Rejection Rate[FRR] : FN / (TN + FN)
- False Acceptance Rate[FAR] : FP / (TP + FP)
- Half Total Error Rate [HTER] : (FAR + FNR) / 2.0
CAISA-MFSD[2012]
- Easy-API[link]
SiW[2018]
- dataset-download-link[passdoor]【磁盘空间约210G!!!】.
- paper : Learning Deep Models for Face Anti-Spoofing: Binary or Auxiliary Supervision[CVPR-2018].
- num_subj = 165【8-live + 20-spoof for each subjec】,total 165 * 28 = 4620 videos.
- name-style : SubjectID_SensorID_TypeID_MediumID_SessionID
- dataset-glimpse