Advanced search Help
Searching for terms: EXACT: "Face Recognition Vendor Test" in: series
Clear all search criteria
Only 2/3! You are seeing results from the Public Collection, not the complete Full Collection. Sign in to search everything (see eligibility).
-
Ongoing Face Recognition Vendor Test (FRVT) Part 5: Face Image Quality Assessment
From the Document: "This report summarizes the ongoing Quality Assessment track of the FRVT [Face Recognition Vendor Test]. Face image quality assessment is a less mature field than face recognition, and so NIST [National Institute of Standards and Technology] regards this work as a development activity rather than an evaluation. In particular, as performance metrics remain under-development - new ones were introduced in this edition of the report - we encourage submission of both new algorithms and comments toward improved formulation and analysis of the problem. Questions, comments and suggestions should be directed to frvt@nist.gov."
National Institute of Standards and Technology (U.S.)
Grother, Patrick J.; Hom, Austin; Ngan, Mei . . .
2021-09-24
-
Face Recognition Vendor Test (FRVT) Part 7: Identification for Paperless Travel and Immigration
From the Executive Summary: "We investigate the use of one-to-many facial recognition in airport transit settings in which travelers' faces are matched against galleries of individuals expected to be present. We primarily consider the case where face recognition serves double-duty for access control (to an aircraft) and facilitation (of recording a visa-holder's departure from a country). This is done in a paperless mode in which a boarding pass (something you have) is replaced with presentation of a biometric (something you are) to a camera, representing an implicit claim to be entitled to board. We describe how such systems can fail, discussing errors during gallery creation, photo capture at boarding, attack detection, and face matching. We discuss how errors might be estimated, citing relevant standards, and their consequences. We quantify face matching errors by simulating departing flights, populating galleries with an airport ENTRY photo of 420 travelers, then measuring accuracy by running searches of EXIT photos. We repeat this with galleries populated with multiple photos per person, and with galleries as large as 42000, modelling the same concept of operations but at a centralized airport checkpoint. We report that accuracy varies greatly across algorithms, that use of multiple images per person reduces errors considerably, and that error rates when searching 42000-person galleries are often three times higher than in 420-person galleries, but still sometimes below 1%. We consider demographics, and note that for the more accurate algorithms, error rates are so low that accuracy variations across sex and race are insignificant. We include additionally a discussion of how our accuracy estimates might differ from those measured operationally due to by factors that we could not control, such as camera type and imaging environment."
National Institute of Standards and Technology (U.S.)
Ngan, Mei; Grother, Patrick J.; Hanaoka, Kayee . . .
2021-07-12
-
Ongoing Face Recognition Vendor Test (FRVT) Part 6B: Face Recognition Accuracy with Face Masks Using Post-COVID-19 Algorithms
From the Executive Summary: "This is the second of a series of reports on the performance of face recognition algorithms on faces occluded by protective face masks commonly worn to reduce inhalation and exhalation of viruses. Inspired by the COVID-19 [coronavirus disease 2019] pandemic response, this is a continuous study being run under the Ongoing Face Recognition Vendor Test (FRVT) executed by the National Institute of Standards and Technology (NIST). In our first report, we tested 'pre-pandemic' algorithms that were already submitted to FRVT 1:1 prior to mid-March 2020. This report augments its predecessor with results for more recent algorithms provided to NIST after mid-March 2020. While we do not have information on whether or not a particular algorithm was designed with face coverings in mind, the results show evidence that a number of developers have adapted their algorithms to support face recognition on subjects potentially wearing face masks. The algorithms tested were one-to-one algorithms submitted to the FRVT 1:1 Verification track. Future editions of this document will also report accuracy of one-to-many algorithms. [...] This report includes[:] [1] Results from evaluating 65 face recognition algorithms provided to NIST since mid-March 2020; [2] Assessment of when both the enrollment and verification images are masked (in addition to when only the verification image is masked); [3] Results for red and white colored masks (in addition to light-blue and black); [4] Cumulative results for 152 algorithms evaluated to date (submitted both prior to and after mid-March 2020)[.]"
National Institute of Standards and Technology (U.S.)
Ngan, Mei; Grother, Patrick J.; Hanaoka, Kayee
2020-11
-
Ongoing Face Recognition Vendor Test (FRVT) Part 1: Verification
From the Document: "'This report' is a draft NIST [National Institute of Standards and Technology] Interagency Report, and is open for comment. It is the sixteenth edition of the report since the first was published in June 2017. Prior editions of this report are maintained on the FRVT [Face Recognition Vendor Test] website, and may contain useful information about older algorithms and datasets no longer used in FRVT."
National Institute of Standards and Technology (U.S.)
Grother, Patrick J.; Ngan, Mei; Hanaoka, Kayee
2019-11-19
-
Face Recognition Vendor Test (FRVT) Part 2: Identification
From the Executive Summary: "This report updates and extends NIST [National Institute of Standards and Technology] Interagency Report 8238 [hyperlink], documenting the evaluation of automated face recognition algorithms submitted to NIST in November 2018. The algorithms, which implement one-to-many identification of faces appearing in two-dimensional images, are prototypes from the research and development laboratories of mostly commercial suppliers, and are submitted to NIST as compiled black-box libraries implementing a NIST-specified C++ test interface. The report therefore does not describe how algorithms operate. The evaluation used four datasets - frontal mugshots, profile views, webcam photos and wild images - and the report lists accuracy results alongside developer names. It will therefore be useful for comparison of face recognition algorithms and assessment of absolute capability. The primary dataset is comprised of 26.6 million reasonably well-controlled live portrait photos of 12.3 million individuals. The three smaller datasets contain more unconstrained photos: 3.2 million webcam images; 200 thousand side-view images; and 2.5 million photojournalism and amateur photographer photos. These datasets are sequestered at NIST, meaning that developers do not have access to them for training or testing. The last dataset, however, consists of images drawn from the internet for testing purposes so while it is not truly sequestered, its composition is unknown to the developers."
National Institute of Standards and Technology (U.S.)
Grother, Patrick J.; Ngan, Mei; Hanaoka, Kayee
2019-09
1
