UniDataPro/ibeta-level-2-certification

iBeta Level 2 Dataset: Presentation Attacks Liveness Detection & Verifications

The dataset consists of more than 32,300 spoofing attacks of 6 different types specifically curated for a passing iBeta Level 2 and getting a certification. It is compliant with the ISO 30107-3 standard, which sets the highest quality requirements for biometric testing and attack detection solutions.

By geting the iBeta Level 2 certification, biometric technology companies demonstrate their commitment to developing robust and reliable biometric systems that can effectively detect and prevent fraud - Get the data

Attacks in the dataset

This dataset is designed to evaluate the performance of biometric authentication and identity recognition technology in detecting presentation attacks, it includes different pad tests.

Videos in the dataset:

1. Real Person: real videos of people
2. 2D Mask: printed photos of people's faces cut out along the contour
3. 2D Mask with Eyeholes: printed photos of people with holes for eyes
4. Latex Mask: latex masks on people
5. Wrapped 3D Mask: 3D cardboard mask attached to a mannequin
6. Silicone Mask: silicone masks on people

💵 Buy the Dataset: This is a limited preview of the data. To access the full dataset, please contact us at https://unidata.pro to discuss your requirements and pricing options.

Metadata for the dataset

Devices: Mi10s, Google Pixel 4, Samsung Galaxy A03s, iPhone 11, iPhone SE 2

Resolution: 1920 x 1080 and higher

The iBeta Level 2 dataset is an essential tool for the biometrics industry, as it helps to ensure that biometric systems meet the highest standards of anti-spoofing technology. This dataset is used by various biometric companies in various applications and products to test and improve their biometric authentication solutions, face recognition systems and facial liveness detection methods.

Frequently Asked Questions

Who can benefit from this iBeta Level 2 Certification Dataset?

This iBeta dataset can benefit biometric security researchers, computer vision engineers, liveness detection developers, identity verification providers, fintech companies, and teams developing presentation attack detection systems. It is particularly relevant for organizations evaluating whether facial recognition systems can distinguish genuine users from physical spoofing attempts.

What video quality is available for liveness detection experiments?

The recordings range from 1920 × 1080 to 3840 × 2160 pixels, covering Full HD through 4K resolution. This gives you enough spatial detail to investigate visual characteristics associated with presentation attacks, including facial contours, mask boundaries, texture, and lighting behavior.

How can the nine backgrounds improve robustness testing?

The recordings were made against nine different backgrounds, introducing contextual variation into the biometric dataset. This is useful because a model can unintentionally learn environmental shortcuts instead of features associated with presentation attacks.

🌐 UniData provides high-quality datasets, content moderation, data collection and annotation for your AI/ML projects