VERA FingerVein

VERA FingerVein is a dataset for fingervein recognition and fingervein presentation attack detection (anti-spoofing). The dataset consists of 440 NIR Bona-Fide images from 110 clients captured using an Open Sensor. The dataset also contains presentation attacks (spoofing attacks) to the same 440 images that can be used to study vulnerability of vein recognition systems or to develop presentation attack detection techniques.

Data

All fingervein samples have been recorded using the open finger vein sensor described in [BT12]. A total of 110 subjects presented their 2 indexes to the sensor in a single session and recorded 2 samples per finger with 5 minutes separation between the 2 trials. The database, therefore, contains a total of 440 samples and 220 unique fingers.

The recordings were performed at 2 different locations, always inside buildings with normal light conditions. The data for the first 78 subjects derives from the the first location while the remaining 32 come from the second location.

The dataset is composed of 40 women and 70 men whose ages are between 18 and 60 with an average at 33. Information about gender and age of subjects are provided with our dataset interface.

Samples are stored as follow with the following filename convention: full/bf/004-F/004_L_2. The fields can be interpreted as <size>/<source>/<subject-id>-<gender>/<subject-id>_<side>_<trial>. The <size> represents one of two options full or cropped. The images in the full directory contain the full image produced by the sensor. The images in the cropped directory represent pre-cropped region-of-interests (RoI) which can be directly used for feature extraction without region-of-interest detection. We provide both verification and presentation-attack detection protocols for full or cropped versions of the images.

The <source> field may one of bf (bona fide) or pa (presentation attack) and represent the genuiness of the image. Naturally, biometric recognition uses only images of the bf folder for all protocols as indicated below. The <subject-id> is a 3 digits number that stands for the subject's unique identifier. The <gender> value can be either M (male) or F (female). The <side> corresponds to the index finger side and can be set to either "R" or "L" ("Right" or "Left"). The <trial> corresponds to either the first (1) or the second (2) time the subject interacted with the device.

Images in the full folder are stored in PNG format, with a size of 250x665 pixels (height, width). Size of the files is around 80 kbytes per sample.

Biometric Data Acquisition Protocol

Subjects were asked to put their index in the sensor and then adjust the position such that the finger is on the center of the image. Bram Ton's Graphical User Interface (GUI) was used for visual feedback, Near Infra Red light control and acquisition. When the automated light control was performing unproperly the operator adjusted manually the intensities of the leds to achieve a better contrast of the vein pattern.

Subjects first presented an index, then the other, a second time the first index and a second time the second index. The whole process took around 5 minutes per subject in average.

The file metadata.csv contains additional information of gender and age (at the time of capture) for each of the 110 individuals available in the dataset.

Presentation-Attack Protocol

To create effective presentation attacks for this dataset, images available were printed on high-quality (200 grams-per-square-meter - GSM) white paper using a laser printer (toner can absorb to near-infrared light used in fingervein sensors), and presented to the same sensor. More information and details can be found on Section 2.2 of the original publication [TVM14].

All presentation attacks were recorded using the same open finger vein sensor used to record the Biometric Recognition counterpart. Images are stored in PNG format, with a size of 250x665 pixels (height, width). Files are named in a matching convention to their counterparts in the biometric recognition. Size of the files is around 80 kbytes per sample.

All bonafide samples corresponds to unaltered originals from the bf part of the dataset.

Images in the full folder are stored in PNG format, with a size of 250x665 pixels (height, width). Size of the files is around 80 kbytes per sample.

Region-of-Interest Annotations

This repository contains the annotations for the fingervein recognition "VERA" fingervein dataset. Each annotation is a text file with points which mark the region-of-interest (RoI) in each image, containing the finger region and excluding the background. To make use of the annotations, you must join the points creating a polygon.

Each annotation file contains annotation for a single, matching image in the original raw dataset. Each file is composed of as many lines as points annotated. There isn't a fixed number of annotations per file. The number of annoated points depends only on the finger contour - some fingers will have therefore more annotations than others. Each point is represented as two (16-bit) unsigned integer numbers representing the y and x coordinates in this order.

Annotations cover only raw data in the full dataset directory. We don't provide annotations for the data in the cropped directory for obvious reasons.

References

If you use this database in your publication, please cite the following publication:

Pedro Tome, Ramachandra Raghavendra, Christoph Busch, Santosh Tirunagari, Norman Poh, B. H. Shekar, Diego Gragnaniello, Carlo Sansone, Luisa Verdoliva and Sébastien Marcel : "The 1st Competition on Counter Measures to Finger Vein Spoofing Attacks", in Proceedings of The 8th IAPR International Conference on Biometrics (ICB), 2015
10.1109/ICB.2015.7139067
http://publications.idiap.ch/index.php/publications/show/3095