FRVT 1:1 [Latest Report] [Agreement] [API] [Validation] [Encryption] [FAQs]
FRVT 1:N [Homepage] [API]
NIST will be accepting submissions again for Ongoing FRVT 1:1 between October 1 - 19, 2018.
The Latest Report includes results for thirty eight algorithms applied to six datasets. The report, which is a draft and open for comments, will be updated on a monthly basis as algorithms and computations complete, as datasets are added, and as new analyses are developed.
To receive update notifications, subscribe by email via frvt-news-request [at] nist.gov (subject: subscribe) (subscribe).
The leaderboard shows the top performing algorithms measured on false non-match rate (FNMR) across several different datasets. FNMR is the proportion of mated comparisons below a threshold set to achieve the false match rate (FMR) given in the header on the fourth row. FMR is the proportion of impostor comparisons at or above that threshold.
# | Developer | VISA Photos FNMR@ FMR ≤ 1e-06 |
VISA Photos FNMR@ FMR ≤ 0.0001 |
MUGSHOT Photos FNMR@ FMR ≤ 0.0001 |
WILD Photos FNMR@ FMR ≤ 0.0001 |
CHILD EXP Photos FNMR@ FMR ≤ 0.01 |
Submission Date |
---|---|---|---|---|---|---|---|
1 | yitu-000 | 0.0331 | 0.0217 | 0.0171 | 0.4315 | 0.5867 | 2017_05_23 |
2 | vocord-002 | 0.0342 | 0.0131 | 0.0192 | 0.94839 | 0.76222 | 2017_06_07 |
3 | visionlabs-002 | 0.0513 | 0.0142 | 0.0203 | 0.3973 | 0.5565 | 2017_09_08 |
4 | fdu-000 | 0.0584 | 0.0163 | 0.0214 | 0.5499 | -41 | 2017_11_22 |
5 | ntechlab-002 | 0.0655 | 0.0215 | 0.0236 | 0.3242 | 0.4591 | 2017_08_23 |
6 | tongyitrans-002 | 0.0666 | 0.03010 | 0.03917 | 0.72523 | 0.74617 | 2017_07_15 |
7 | tongyitrans-001 | 0.0727 | 0.03819 | 0.04118 | 0.70420 | 0.74316 | 2017_04_01 |
8 | ntechlab-001 | 0.0838 | 0.0258 | 0.03011 | 0.3191 | 0.4722 | 2017_05_10 |
9 | tevian-000 | 0.1299 | 0.03617 | 0.0225 | 0.5008 | -40 | 2017_12_10 |
10 | morpho-000 | 0.13410 | 0.0269 | 0.0289 | 0.89336 | 0.84633 | 2017_07_11 |
11 | vocord-001 | 0.14111 | 0.03515 | 0.06330 | 0.65414 | 0.69511 | 2017_04_21 |
12 | 3divi-002 | 0.15412 | 0.0216 | 0.03315 | 0.4957 | -39 | 2017_10_20 |
13 | 3divi-001 | 0.15413 | 0.0204 | 0.03012 | 0.4926 | 0.5494 | 2017_06_22 |
14 | yisheng-001 | 0.16014 | 0.03212 | 0.04829 | 0.62811 | 0.75620 | 2017_08_22 |
15 | neurotechnology-002 | 0.16615 | 0.03616 | 0.04119 | 0.4274 | 0.6838 | 2017_11_02 |
16 | visionlabs-001 | 0.18016 | 0.03011 | 0.0247 | 0.59110 | 0.5616 | 2017_06_12 |
17 | innovatrics-001 | 0.18317 | 0.03413 | 0.04322 | 0.64313 | 0.81530 | 2017_07_31 |
18 | rankone-003 | 0.18418 | 0.03820 | 0.0288 | 0.67415 | -37 | 2017_11_28 |
19 | innovatrics-000 | 0.19119 | 0.03414 | 0.04627 | 0.72022 | 0.85234 | 2017_07_31 |
20 | yisheng-000 | 0.19920 | 0.03718 | 0.02910 | 0.68618 | 0.74918 | 2017_08_22 |
Additional algorithms not listed in the leaderboard can be found in Table 2 of our latest FRVT report.
Prior editions of the report: 2017-10-03 | 2017-08-25 | 2017-08-07 | 2017-07-31 | 2015-05-15 | 2017-05-01 | 2017-04-12 | 2017-04-03 | 2017-03-23.
Developers may submit their algorithms at any time by following instructions below. Results will be posted as soon as algorithms successfully complete the tests.
NIST is starting a new evaluation of face recognition technologies starting in February 2017. Unlike our previous evaluations, this activity will be conducted on an ongoing basis: The evaluation will remain open indefinitely such that developers may submit their algorithms to NIST whenever they are ready, but no more frequently than three calendar months. The algorithms will be evaluated rapidly on a first-come-first-served basis, following our MINEX III evaluation of fingerprint recognition implementations. Performance results will be posted to the NIST website as soon as they are ready. This approach more closely aligns evaluation with development schedules; this improves over the two to four year interval between past FRVT tests.
The FRVT is aimed at measurement of the performance of automated face recognition technologies applied to a wide range of civil, law enforcement and homeland security applications including verification of visa images, de-duplication of passports, recognition across photojournalism images, and identification of child exploitation victims. In all cases the input image will contain one face only. Our performance reports will include measurements of accuracy, speed, storage and memory consumption, and resilience. NIST will report the dependence of performance on the properties of the images and the subjects. In its initial form, FRVT has one assessment track, for face verification.
FAQs [last update: 2017-01-30]: Ongoing responses to a number of questions regarding the evaluation are addressed in our FAQs document.
Important Dates [last update: 2018-09-26]: The submission window for Ongoing FRVT 1:1 will reopen from October 1 - 19, 2018.
Inquiries and comments may be submitted to frvt [at] nist.gov (frvt[at]nist[dot]gov).
frvt-news-request [at] nist.gov (subject: subscribe) (Subscribe) to the FRVT mailing list to receive emails when announcements or updates are made.