New ELFT Dataset: IARPA N2N (Sequestered)

12 views

Skip to first unread message

Evaluation of Latent Friction Ridge Technology (ELFT)

unread,

Mar 9, 2026, 4:40:00 PMMar 9

to Evaluation of Latent Friction Ridge Technology (ELFT)

Hello ELFT stakeholders,

We’ve added a new dataset, IARPA N2N (Sequestered), for all active ELFT algorithms. All report cards have been updated, and new columns have been added to the results table. Algorithms were run under Ubuntu 24.04.

IARPA N2N (Sequestered) features 2000+ latent impression probes that have never been publicly released. Images were collected during the IARPA N2N Fingerprint Capture Challenge, which also led to the creation of the research database NIST Special Database (SD) 302. Importantly, the dataset used in ELFT contains only subjects that were not part of SD 302.

Certified Latent Print Examiners annotated features in all images, though we have only posted results when searching the image without features (so far). The probes in this dataset were deemed “of value for identification” by the CLPEs. Areas outside the CLPE-annotated regions of interest were not masked in white pixels, as is the case in the existing version of SD 302.

In this update, we have also removed placeholders for data related to our previous Michigan State Police dataset, as it will not be returning to the evaluation.