iBeta Level 2 Dataset

There are 25,000+ videos tailored for iBeta level 2 certification from 150+ IDs

Check samples on Kaggle

iBeta Level 2 dataset summary

Parameter	Value
Volume	25,000+ videos from 150+ IDs
Coverage	Silicone, latex, wrapped 3D paper, advanced paper, and cloth 3D mask attacks
Demographics	Adults aged 18–65, balanced gender, multi-ethnic
Devices	iOS and Android phones (10+ device models)
Conditions	Indoor and outdoor, multiple lighting levels and view angles

Introduction

iBeta Level 2 PAD is a compact training dataset for liveness detection and anti-spoofing focused on 3D mask attacks and active liveness. It includes 25,000+ short multi-frame videos captured on diverse iOS and Android devices, subjects, and conditions. Covered attack types: silicone, latex, wrapped 3D paper, advanced paper, and cloth 3D masks. Ideal for training, validation, and pre-certification experiments targeting iBeta Level 2 certification, which tests biometric systems against ISO/IEC 30107-3 – the international standard for biometric presentation attack detection

What Is iBeta Level 2 PAD Certification?

iBeta Level 2 is the second-tier presentation attack detection (PAD) certification administered by iBeta Quality Assurance, an independent NIST-NVLAP-accredited testing laboratory. Level 2 testing goes beyond basic 2D paper and replay attacks evaluated at Level 1, it specifically targets high-realism 3D mask presentation attacks, including silicone masks, latex masks, wrapped 3D paper masks, and advanced paper-based attacks with eyeholes for live blinking

Compliance with iBeta Level 2 is performed under the ISO/IEC 30107-3 international standard and is required for face recognition systems deployed in higher-security applications such as banking, fintech identity verification, government ID, and regulated eKYC. Robust face anti-spoofing models must demonstrate low APCER (Attack Presentation Classification Error Rate) and BPCER (Bona fide Presentation Classification Error Rate) across all Level 2 attack categories to pass iBeta certification

This dataset is purpose-built to provide the training data needed to prepare biometric face recognition and liveness detection systems for iBeta Level 2 PAD evaluation

Dataset Features

Mask dynamics in active liveness: Zoom-in, zoom-out, head turns, and natural blinking
Off-axis viewing angles: Subjects recorded from multiple angles and distances
Dual-device per attack: Every mask attack captured on both iOS and Android phones
Variable accessories: Different hairstyles, glasses, and wigs across sessions

Source and collection methodology

We captured realistic iBeta Level 2 spoofing scenarios with front-facing cameras across varied people, environments, and devices. Each clip follows an active liveness script (zoom-in/zoom-out, natural head turns/blinks) and lasts ~10 seconds

Recording devices
- iOS: iPhone 14, iPhone 14 Pro, iPhone 13 Pro
- Android: Galaxy S23, Xiaomi Redmi Note 12 Pro+, Galaxy A54, Pixel 7, Honor 70
Capture protocol
- 3D mask attacks
- Guided zoom phases
- Multiple distances
Environments & lighting
- Indoor (offices, home settings) and outdoor scenes
- Three lighting levels: low, medium, bright; mixed backgrounds

Real-World Validation: Open Liveness Model Stress Test

To demonstrate the practical value of this dataset, we tested its 3D silicone mask attack samples against Doubango’s open-source face liveness model – a publicly available liveness detection SDK used as a reference implementation by anti-spoofing researchers and developers

Key result: Doubango’s liveness classifier rated both silicone mask attacks below as 99.91% and 99.97% genuine, fully classifying the spoofs as real, living faces

Academic Reference

The canonical academic benchmarks for iBeta Level 2 attack vectors include the Idiap 3DMAD (3D Mask Attack Database) for paper-based 3D masks and the Idiap CSMAD (Custom Silicone Mask Attack Database) for silicone mask attacks – both foundational research datasets from the Idiap Research Institute in face anti-spoofing literature. This commercial dataset extends those research lines with significantly more participants, modern smartphone capture conditions (iOS and Android dual-device), broader demographic representation (Caucasian, Black, Asian, Latinx), active liveness sequences (zoom-in/zoom-out, micro-movements), and direct alignment with iBeta certification protocols, designed for production face recognition systems rather than research benchmarks alone

Training Anti-Spoofing Models with iBeta Level 2 Data

The iBeta Level 2 Dataset extends anti-spoofing model training to 3D attack vectors required for advanced PAD certification. Use this data to train liveness detection models against silicone masks, latex masks, wrapped 3D paper masks, and other sophisticated presentation attacks. The dataset structure matches the iBeta Level 2 test protocol distribution and integrates into existing biometric model training pipelines

Use cases and applications

iBeta Level 2 Certification Compliance:

Helps to train the models for iBeta level 2 certification tests
Allows pre-certification testing to assess system performance before submission

Inhouse Liveness Detection Models:

Used for training and validation of anti-spoofing models
Enables testing of existing algorithms and identification of their vulnerabilities against spoofing attacks

How companies achieved iBeta with us

Datasets like this contributed to 21% of companies that passed iBeta certification in 2025 – all Axon Labs clients

From Failed Attempt to Success: UK Fintech Passed iBeta 2

UK fintech invested in datasets from another vendor for iBeta 2 and failed. The gap analysis revealed critical missing attack vectors. They needed a solution fast

Double Success: E-KYC Startup Achieves iBeta 1 & 2

A US-based e-KYC startup needed to prove their facial recognition technology met the highest industry security standards. Their ambitious goal: pass iBeta 1 and 2 Certification

Technology company from Vietnam: iBeta Level 2 success

A Vietnam-based AI/Big Data firm coached by Axon Labs passed iBeta PAD Level 2 on the first try with 0% successful spoofs; the solution claims 99.9% face-recognition accuracy

Format: Videos are optimized for compatibility with mainstream ML frameworks
Resolution and frame rate: Videos are high-resolution with frame rates calibrated for capturing quick and realistic mask placements, ensuring precise data for model training

Potential customisation options:

Filming videos attacks with targeted movements
Filming videos attacks for you on target devices (for example, webcams)
Using your SDK for custom attack scenarios spoofing your ML model
Use RGB and USB cameras to support diverse research and testing needs
For two masks, video recordings are available from the back camera, capturing multiple angles (close-up, far, left, and right)

Legal & Compliance

We prioritize data privacy, ethical AI development, and regulatory compliance. Our iBeta Level 2 Dataset is collected and processed in full accordance with global data protection standards including GDPR, ensuring legality, security, and responsible AI practices

Sample dataset

A sample version of this dataset is available on Kaggle. Leave a request for additional samples in the form below

Have a question?

What attack types does the iBeta Level 2 Dataset cover?

The iBeta Level 2 Dataset covers the full range of 3D presentation attacks tested at iBeta Level 2 PAD certification: silicone masks, latex masks, wrapped 3D paper masks, advanced paper masks, and cloth 3D masks. This single dataset consolidates all five 3D mask attack types into one training resource, eliminating the need to source multiple 3D mask datasets separately

What is included in the iBeta Level 2 Dataset?

The dataset contains 25,000+ videos from 150+ participants across five distinct 3D mask attack types. Each video is captured on both iOS and Android phones (8 device models total), follows an active liveness protocol with mask dynamics (zoom-in, zoom-out, mask approach and retreat sequences), and covers indoor and outdoor environments across multiple lighting levels and off-axis viewing angles

What iBeta certification level does this dataset prepare models for?

The dataset targets iBeta Level 2 PAD certification - the middle tier of biometric presentation attack detection testing, focused on 3D mask attacks. iBeta Level 2 tests biometric systems against ISO/IEC 30107-3 Level 2 attack instruments, and is the required certification tier for fintech, KYC, and identity verification providers seeking protection against 3D spoofing attempts

How is the iBeta Level 2 Dataset different from Level 1 and Level 3?

iBeta Level 1 covers 2D attacks (paper prints, display replays), easier to defend against but far more common in real-world fraud. Level 2 covers 3D mask attacks (silicone, latex, wrapped 3D paper, advanced paper, cloth 3D), harder to detect and critical for production-grade fraud prevention. iBeta Level 3 focuses only on the highest-fidelity custom-manufactured masks (rubber and 3D resin). Most fintech and KYC providers certify at Level 2 before advancing to Level 3

How were 3D mask attacks captured in this dataset?

3D mask attacks were captured using front-facing (selfie) cameras on 8 iOS and Android device models: iPhone 14, iPhone 14 Pro, iPhone 13 Pro, Galaxy S23, Xiaomi Redmi Note 12 Pro+, Galaxy A54, Pixel 7, and Honor 70. Each mask attack follows an active liveness protocol including zoom-in/out, head turns, natural blinking, and mask approach/retreat sequences. Captures span indoor and outdoor environments across three lighting tiers (low, medium, bright) and multiple viewing angles

Has this dataset been proven effective for iBeta certification preparation?

Yes, with two types of validation. We stress-tested silicone mask samples against Doubango's open-source face liveness model, which rated the attacks at 99.91% and 99.97% genuine, confirming the dataset produces spoofs that bypass respected liveness detection implementations. In addition, 21% of companies that passed iBeta certification in 2025 are Axon Labs clients, including a UK fintech that passed iBeta 2 after a failed attempt with a competitor's dataset, a US-based e-KYC startup that achieved both iBeta 1 and 2 certification, and a Vietnamese AI company that passed iBeta PAD Level 2 on first try with 0% successful spoofs

Is the iBeta Level 2 Dataset GDPR-compliant for commercial use?

Yes. All data is collected with explicit written informed consent from every participant in compliance with GDPR Article 9, which specifically governs the processing of biometric data. The dataset is licensed for commercial use in AI model training, validation, and iBeta certification preparation. Comprehensive compliance documentation, including consent provenance, collection methodology, and legal basis review, is available upon request for regulatory or audit purposes

Can I get a sample of the iBeta Level 2 Dataset before purchasing?

Yes. A sample version of this dataset is available on request, you can verify attack diversity across all five mask types, video quality, active liveness sequences, and format compatibility with your training pipeline before committing to the full dataset. Submit a request through the form on this page for sample access, typically delivered within 1–2 business days

Contact us

Tell us about yourself, and get access to free samples of the dataset

I want to receive communications on the newly added datasets

Didn't find what you were looking for?

Our collection includes many datasets for various requests

iBeta

iBeta Level 2 Dataset