Browse › Computer Vision › Face Verification › IJB-C
IJB-C Benchmark (Face Verification)
Face Verification is a machine learning task in computer vision that involves determining whether two facial images belong to the same person or not. The task involves extracting features from the facial images, such as the shape and texture of the face, and then using these features to compare and verify the similarity between the images.
The archive carries no text for this table; the description above is the archive's text for the task Face Verification. archive 2025-07-28
Over time archive 2025-07-28
The chart needs JavaScript; the table below carries every value.
Not inferred: TAR @ FAR=1e-6, TAR @ FAR=1e-5, TAR @ FAR=1e-4, TAR @ FAR=1e-3, TAR @ FAR=1e-2, training dataset, model, Rank-1, Rank-5. Points are placed at the row's paper date; 26 of 26 rows carry one.
Results archive 2025-07-28
Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.
| Paper | Code | Ran Syntology | Report | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | HeadSharing: SH-KD | 90.24% | 93.73% | 95.64% | MS1M V3 | MobileFaceNet | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 2 | HeadSharing: TH-KD | 89.82% | 93.50% | 95.48% | MS1M V3 | MobileFaceNet | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 3 | ArcFace+CSFM | 89.34% | 94.06% | 95.9% | 96.31 | 97.48 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 4 | Partial FC | 97.23% | 98.00% | WebFace42M | ViT-L | – | Paper | Code | 2022 | 4 of 12 ran · 8 unverified | report | |||||
| 5 | PartialFC | 96.93% | 97.97% | WebFace42M | R200 | – | Paper | Code | 2022 | 4 of 12 ran · 8 unverified | report | |||||
| 6 | ArcFace | 96.07% | IBUG-500K | R100 | – | Paper | Code | 2018 | 16 of 21 ran · 5 unverified | report | ||||||
| 7 | Mag+UNPG | 94.7% | – | Paper | Code | 2022 | 2 of 14 ran · 12 unverified | report | ||||||||
| 8 | Cos+UNPG | 94.47% | 96.38% | 97.57 | MS1MV2 | R100 | – | Paper | Code | 2022 | 2 of 14 ran · 12 unverified | report | ||||
| 9 | L2E+IS-sampling | 93.25% | 95.49% | 97.05% | MS1M V3 | MobileFaceNet | – | Paper | – | 2021 | no code linked | report | ||||
| 10 | MagFace++ | 90.36% | 95.97% | MS1MV2 | R100 | – | Paper | Code | 2021 | 1 of 2 ran · 1 unverified | report | |||||
| 11 | circle loss | 89.60% | 93.95% | 96.29% | MS1M Cleaned | R100 | – | Paper | Code | 2020 | 2 of 2 ran · 0 unverified | report | ||||
| 12 | WebFace42M baseline | 97.7% | WebFace42M | R100 | – | Paper | – | 2021 | no code linked | report | ||||||
| 13 | AdaFace (WebFace4M) | 97.39% | – | Paper | Code | 2022 | 17 of 22 ran · 5 unverified | report | ||||||||
| 14 | FFC | 97.31% | WebFace42M | R100 | – | Paper | Code | 2021 | linked, not harvested | report | ||||||
| 15 | CAFace+AdaFace (WebFace4M) | 97.3% | 98.08 | – | Paper | Code | 2022 | 7 of 9 ran · 2 unverified | report | |||||||
| 16 | AdaFace (MS1MV3) | 97.09% | – | Paper | Code | 2022 | 17 of 22 ran · 5 unverified | report | ||||||||
| 17 | AdaFace (MS1MV2) | 96.89% | – | Paper | Code | 2022 | 17 of 22 ran · 5 unverified | report | ||||||||
| 18 | ElasticFace-Cos | 96.57% | MS1M V2 | R100 | – | Paper | Code | 2021 | linked, not harvested | report | ||||||
| 19 | Arc+UNPG | 96.33% | 97.51 | – | Paper | Code | 2022 | 2 of 14 ran · 12 unverified | report | |||||||
| 20 | QMagFace | 96.19% | 97.62 | 98.51 | – | Paper | Code | 2021 | linked, not harvested | report | ||||||
| 21 | CurricularFace | 96.1% | – | Paper | Code | 2020 | 5 of 13 ran · 8 unverified | report | ||||||||
| 22 | PFEfuse + match | 95.49% | 97.17% | MS1M V2 | SphereFace64 | – | Paper | Code | 2019 | 1 of 1 ran · 0 unverified | report | |||||
| 23 | VGGFace2_ft | 92.7% | 96.7% | Vggface2 | R50 | – | Paper | Code | 2017 | 5 of 8 ran · 3 unverified | report | |||||
| 24 | AIM | 93.5% | – | Paper | Code | 2018 | 0 of 6 ran · 6 unverified | report | ||||||||
| 25 | MN-vc | 92.70% | – | Paper | Code | 2018 | 0 of 4 ran · 4 unverified | report | ||||||||
| 26 | FaceNet | 66.5% | – | Paper | Code | 2015 | 65 of 154 ran · 89 unverified | report |
All 26 rows shown. 26 link to a paper page on this site; 0 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28
Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 18 rows have a graph line, from 13 distinct papers; 16 rows (11 papers) have at least one sample that ran. Counting each paper once: Syntology ran 125 of 268 samples; 143 unverified. Separately, 80 of those 268 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-24. Per-sample status is on the paper page.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections