LatentSleuth
For the prints AFIS gave up on.
LatentSleuth is a matching tool built for the cases that pile up in examiner backlogs — palm fragments and difficult finger impressions with no cores, unknown orientation, or minimal friction ridge detail. Designed around how examiners actually work.
The hard impressions don't make it through standard AFIS.
Partial palms and smeared fingers get rejected.
Latent impressions that lack cores, exhibit unknown or misleading orientation, or contain minimal friction ridge detail are exactly the ones automated systems reject — and exactly the ones that pile up in examiner backlogs.
-
Lack of coresStandard matchers anchor on cores; the prints that don't have them don't get matched.
-
Unknown or misleading orientationTips, joints, and rotated fragments confuse minutiae-only algorithms.
-
Minimal friction ridge detailWhen minutiae count is low, AFIS returns no candidates — the examiner is on their own.
Match on every usable signal, not just minutiae.
LatentSleuth 2.0 incorporates all usable information rather than only minutiae as in traditional approaches, so more challenging latent prints can be searched. It offloads complex print searching to a novel matching algorithm purpose-built for difficult evidence.
-
Beyond-minutiae matchingUses all usable information in the impression to outperform existing matchers on difficult latents.
-
Offloads complex searchComplex print searching is handled by the novel matching algorithm, not the examiner.
-
Fits the examiner workflowCompare against suspect references, victim elimination sets, and AFIS candidates — in one tool.
Days worth of work in a matter of minutes.
Distortion Free Overlays
Automatically finds the best location for the Latent ROI in the fingerprint image and creates a distortion free overlay of the Latent ROI onto that location in the fingerprint image
Palm fragments AFIS rejects
LatentSleuth handles partial palms — the fragments AFIS treats as unmatchable. Ridge-skeleton matching is rotationally independent and tolerates the distortion from “elasticity of skin,” so the algorithm can place a fragment against a full reference palm even with few minutiae and no clear orientation.
Accurate overlays for placement
LatentSleuth produces precise overlays of the latent print onto each reference print, eliminating the time the LPE spends searching for the right orientation and placement.
Designed around how examiners actually work.
Revolutionary matching
Advanced algorithms incorporate all usable information in the impression — not just minutiae.
Challenging impressions
Handles palm fragments and difficult finger impressions that other tools struggle with.
AFIS integration
Works alongside existing AFIS workflows and search results — useful for the candidates AFIS returned and the ones it didn't.
Examiner-grade tool
Built specifically for Latent Print Examiners and forensic professionals — not retrofitted from a generic matcher.
Productivity
Off-loading the search on very small latents (lacking orientation or anatomical cues) to the algorithm leaves examiners free for the evaluation phase.
Consistent
The similarity of a candidate Image is measured relative to a base set.
One integration, extending your AFIS on the most difficult problems.
Peer-reviewed research underpinning LatentSleuth.
Published in Forensic Science International and developed in collaboration with the FBI Laboratory Division, George Mason University, and the Virginia Department of Forensic Science.
LatentSleuth: Empowering Latent Print Examiners
Product overview describing LatentSleuth’s ridge-skeleton matching method, rotational independence, accurate overlays for ACE-V, and the ability to match tips and phalangeal areas — not just standard rolls and flats.
Read overview →Statistical Error Estimation for an Objective Measure of Similarity to a Latent Image
Gantz, Miller. Uses LatentSleuth’s WARP technology to define a Hierarchical Median statistic for Level-2 similarity to a latent ROI, then constructs a Null Distribution to predict rarity of outlying reference-image similarities — an objective error measure for LPE casework.
Read poster →A novel approach for latent print identification using accurate overlays to prioritize reference prints
Gantz, Gantz, Walch, Roberts, Buscaglia. On NIST SD27, fully automated overlays placed the true mate reference first for 80.7% of latents and in the top ten for 89.8% — rising to 90.9% and 96.6% after manual post-processing.
Read paper →What teams ask before they pilot.
Does LatentSleuth offer exports that I can use in court?
Can you operate the environment if I don't want to?
How is LatentSleuth priced?
See LatentSleuth on the prints AFIS rejected.
Demos, free trials, and pricing — usually within a week of a first call.