Results


Currently there are only results for the restricted protocol.

See the instructions below on how to generate the ROC curves.
These are the contact details for submitting new results (for accepted papers to a peer reviewed publication) on this dataset.




Restricted Protocol Results:

Initial Results: See paper for explanation on each method.

results Type index:
1. All pairs comparisons. 2. Pose based methods. 3. Algebraic methods. 4. Non-algebraic Set methods. 5. Matched Background Similarity (MBGS).

Selected Results from the above table + results submitted by other groups:

Ref
 Method Accuracy ± SE AUC EER
[1]
   Min dist, FPLBP   65.6 ± 1.8   70.0   35.6
[1]
   Min dist, LBP   65.7 ± 1.7   70.7   35.2
[1]
   ||U1'U2||, FPLBP   64.3 ± 1.6   69.4   35.8
[1]
   ||U1'U2||, LBP   65.4 ± 2.0   69.8   36
[1]
   MBGS L2 mean, FPLBP   72.6 ± 2.0   80.1   27.7
[1]
   MBGS L2 mean, LBP   76.4 ± 1.8   82.6   25.3
[2]
   MBGS+SVM-
   78.9 ±1.9 86.9
 21.2
[3]
   APEM-FUSION
   79.1 ±1.5 86.6
 21.4
[4]
   STFRD+PMML
   79.5 ±2.5 88.6
19.9
[5]
   VSOF+OSS (Adaboost)
   79.7 ±1.8 89.4
20.0
[6]
   DeepFace-single
   91.4 ±1.1 96.3
8.6
[7]
   DDML (LBP)
   81.3 ±1.6 88.7
19.7
[7]
   DDML (combined)
   82.3 ±1.5 90.1
18.5
[8]
   EigenPEP
   84.8 ±1.4 92.6
15.5
[9]
   LM3L
   81.3 ±1.2 89.3
19.7
[10]
   CNN-3DMM estimation (Network model)
   88.80 ±2.21 95.4
11.2
ROC Curves: ROC curve

Generate ROC Curves:

Generate the ROC curve using this gnuplot script: create_roc_curve.p.
The script's input is a text file for each method, where each line represents a point on the ROC curve. The format is:
[average true positive rate] [average false positive rate]

Here are the currently available ROC files:
  MBGS_L2_FPLBP_mean.txt
  MBGS_L2_LBP_mean.txt
  MIN_DIST_FPLBP.txttxt
  MIN_DIST_LBP.txt
  U1U2f_FPLBP.txt
  U1U2f_LBP.txt
  APEM-FUSION.txt
  STFRD+PMML.txt
  VSOF+OSS(Adaboost).txt
  youtube_roc_deepface_single.txt
  ddml_lbp_restricted.txt
  ddml_combined_restricted.txt
  EigenPEP.txt
  lm3l_restricted_ytf.txt
  3DMM_CNN_ytf.txt

References:
[1]  Lior Wolf, Tal Hassner and Itay Maoz. Face Recognition in Unconstrained Videos with Matched Background   Similarity. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2011.
[2]  Lior Wolf and Noga Levy. The SVM-minus Similarity Score for Video Face Recognition. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2013.
[3]  Haoxiang Li, Gang Hua, Zhe Lin, Jonathan Brandt, Jianchao Yang. Probabilistic Elastic Matching for Pose Variant Face Verification. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2013.
[4]  Zhen Cui, Wen Li, Dong Xu, Shiguang Shan and Xilin Chen. Fusing Robust Face Region Descriptors via Multiple Metric Learning for Face Recognition in the Wild. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2013.
[5]  Heydi Mendez-Vazquez, Yoanna Martinez-Diaz, Zhenhua Chai. Volume Structured Ordinal Features with Background Similarity Measure for Video Face Recognition. International Conference on Biometrics (ICB), 2013.
[6]  Yaniv Taigman, Ming Yang, Marc'Aurelio Ranzato, Lior Wolf. DeepFace: Closing the Gap to Human-Level Performance in Face Verification. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2014.
[7]  Junlin Hu, Jiwen Lu, and Yap-Peng Tan. Discriminative Deep Metric Learning for Face Verification in the Wild. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2014.
[8]  Haoxiang Li, Gang Hua, Xiaohui Shen, Zhe Lin, and Jonathan Brandt. Eigen-PEP for Video Face Recognition. The 12th Asian Conference on Computer Vision (ACCV), 2014.
[9]  Junlin Hu, Jiwen Lu, Junsong Yuan, Yap-Peng Tan. Large Margin Multi-Metric Learning for Face and Kinship Verification in the Wild. The 12th Asian Conference on Computer Vision (ACCV), 2014.
[10]  Anh Tuan Tran, Tal Hassner, Iacopo Masi and Gerard Medioni, "Regressing Robust and Discriminative 3D Morphable Models with a very Deep Neural Network," arXiv preprint arXiv:1612.04904, 15 Dec. 2016. Project page