CANDY project mints sweet vein biometric dataset

A team of researchers has used a new contactless biometric sensor platform for vascular and surface biometrics, such as both palm veins and palm prints, to collect a new dataset for research into vein biometrics accuracy and fraud attacks.

The sensor platform developed by members of the Idiap Research Institute in collaboration with Global ID is called sweet. It supports biometric data acquisition through multi-spectral near-infrared (NIR), RGB-color, stereo vision and photometric stereo scanning. “An Open Source Modular Platform for Contactless Hand Vascular Biometric Experiments” is described in a research paper.

Using the platform, researchers found error rates matching single finger-vein biometrics of left hands of 3.5 percent to 5 percent. Right-hand finger-vein match rates were “very high,” but in both cases, a score-fusion of three fingers reducers classification errors.

The platform and dataset are a product of the Innosuisse CANDY project, which has received funding from the Innosuisse agency and the Swiss Center for Biometrics Research and Testing. CANDY is an acronym derived from “Contactless finger vein recognition and presentation attack detection on-the-fly.”

The goal of the CANDY project is to reliably identify people in medical settings without having to clean scanners of potential contamination, initially inspired by COVID-19. The researchers also see potential application in commercial point-of-sale systems, such as for biometric retail purchases.

The new dataset, described in a separate paper titled “Vascular Biometrics Experiments on Candy — A New Contactless Finger-Vein Dataset,” consists of finger, palm and wrist biometric data from 120 subjects, plus 280 finger-vein presentation attack instruments (PAIs). The dataset contains 1,200 genuine vein biometrics samples, while the PAIs were used to create 1,400 attack samples, and consist of 14 species. The researchers call the dataset “Candy_v3.”

The paper will be presented at the IEEE International Conference on Pattern Recognition (ICPR) 2024.

Global ID also participated in a recent gathering of biometrics and digital identity technology providers to exchange ideas at a Sopra Steria facility, along with Thales and Idemia. The vein biometrics company praised Sopra Steria and Idiap Senior Researcher Sébastien Marcel for leading the collaborations.

On the practical side, Global ID’s finger-vein biometrics have been deployed to Namibia to help reduce HIV infections among vulnerable youth in Project HOPE.

Article Topics

biometrics  |  CANDY project  |  dataset  |  finger vein  |  Global ID  |  Idiap  |  palm vein biometrics  |  Swiss Center for Biometrics Research and Testing  |  vein biometrics