iBeta Level 1 Dataset
Check samples on Kaggle
iBeta Level 1 dataset summary
|
Parameter
|
Value
|
|---|---|
|
Volume
|
Paper attacks: 22,000+ videos from 80+ participants
Replay attacks: 8,000+ videos from 2,500+ participants |
|
Coverage
|
Paper, cutout, and replay attacks across iBeta Level 1 vectors
|
|
Demographics
|
Adults aged 18–65, balanced gender, multi-ethnic
|
|
Devices
|
iOS and Android phones (multiple device models)
|
|
Conditions
|
Indoor and outdoor, varied lighting and backgrounds
|
Introduction
The iBeta Level 1 Paper & Replay Attacks Dataset offers a comprehensive collection of presentation attacks (PAD) tailored for iBeta Level 1 testing. Beyond paper-based masks and printouts, it includes a diverse set of replay attacks, photo and video replays on smartphone and laptop displays under varying brightness levels, distances, and angles: to reflect real-world spoofing scenarios. Designed for researchers and developers working on liveness detection, this dataset provides broad coverage for training and validating anti-spoofing models, delivering end-to-end completeness for iBeta Level 1 certification, which tests biometric systems against ISO/IEC 30107-3 – the international standard for biometric presentation attack detection
Dataset Features
- Active liveness sequences: Zoom-in, zoom-out, head turns, and natural blinking
- Replay variation axes: Multiple brightness levels, distances, and viewing angles
- Multi-display replay capture: Mobile, laptop, and PC monitors used as replay surfaces
Source and collection methodology
Data were collected from real-life selfies and short videos provided by participants, followed by two families of presentation attacks
- Paper attacks: print, cutout, cylinder, and 3D mask variations, recorded on both iOS and Android with controlled changes in angle, distance, and lighting
- Replay attacks (mobile + PC): photos/videos of the same participants replayed on smartphone screens (iOS/Android) and desktop monitors. Replay clips (~5–12+ s) include slow camera motion, zoom-in/zoom-out phases, varied brightness, viewing angles, and distances; phone borders are hidden when applicable
All sequences contain explicit zoom-in and zoom-out segments to support active liveness detection and to simulate realistic spoofing attempts
This dataset provides 5 variations of spoof attacks
Some of the spoof attacks in our dataset were tested on Doubango, a leading 3D liveness detection framework
Doubango performs advanced 3D liveness checks using a single 2D image and claims to outperform market leaders like FaceTEC, BioID, Onfido, and Huawei in both speed and accuracy
During testing, our attack images bypassed Doubango’s security checks, with the system generating green bounding boxes around the faces (indicating acceptance as “live” users). This confirms that the attacks were not flagged as spoofs, demonstrating their ability to trick even high-performance systems
These results highlight the quality of our dataset for training robust anti-spoofing models capable of defending against evolving threats in real-world scenarios
1. Real life selfie & videos from participants
Genuine facial data collected in various lighting conditions and angles to ensure robust system evaluation
2. Print and Cutout paper attacks
Attackers use printed photos or cutout masks with eye mouth holes to trick recognition systems
3. Cylinder attack to create volume effect
A printed face is wrapped around a cylindrical object to simulate a 3D structure. This method is effective in deceiving simple 2D detection algorithms
4. Paper attacks on Actor with head/eyes variations
A paper face is placed over a real person’s head to mimc real facial movement. Variation include blinking, head tilts, and expressions to test system resilience
5. 3D paper masks with volume based elements such as nose
High-quality 3D masks icorporate raised features sucj as a nose to enhance realism. More challenging for liveness detection algorithms
5. PC/Mobile Replay attacks
A pre-recorded video of a real face is played on a phone, or laptop screen and captured by the camera as if it were a live user. Variations include different angles, distances, screen brightness, and glare to account for screen quality and reflections
Why Axonlabs better than competitors
One of our partners tested our dataset and a competitor’s dataset using their own liveness detection model while preparing for iBeta Level 1 certification. The results show a clear difference in difficulty between the two datasets. Both datasets were tested on a sample of approximately 200 attack attempts each, ensuring a fair comparison
• Our dataset presents a greater challenge for liveness detection models. The model frequently misclassified attack images as real (label 0), meaning our spoofing techniques are more advanced and harder to detect
• The competitor’s dataset, on the other hand, was mostly detected as attacks (score 1), except for a single type of attack where the model showed some uncertainty
This demonstrates that our dataset provides more value for training robust liveness detection models, as it exposes them to more deceptive and realistic attacks
Understanding the score:
Horizontal axis: score value (0 – model judges the frame as “live”, 1 – “spoof”). Dot color shows ground truth: green = genuine face, red = spoof attack
By training on a more challenging dataset, models can significantly improve their spoof detection capabilities, making them more resilient against real-world threats
Use cases and applications
iBeta Level 1 Certification Compliance:
- Helps to train the models for iBeta level 1 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
Legal & Compliance
We prioritize data privacy, ethical AI development, and regulatory compliance. Our Silicone Mask Attack Dataset is collected and processed in full accordance with global data protection standards including GDPR, ensuring legality, security, and responsible AI practices
Download information
Sample data are available on Kaggle as three separate datasets: Paper Attacks (sample), Replay Attacks — Mobile (sample), and Replay Attacks — PC/Laptop (sample). Request full access or additional samples via the form below
Have a question?
The iBeta Level 1 Dataset covers the full range of 2D presentation attacks tested at iBeta Level 1 PAD certification: photo print attacks, cutout print attacks, cylinder-mounted prints, 3D paper masks, and display replay attacks across smartphones, laptops, and PC monitors. This single dataset covers all this types of attacks
The dataset combines two distinct sub-collections: 22,000+ paper attack videos from 80+ participants and 8,000+ display replay attack videos from 2,500+ participants. Each video is captured in front-facing (selfie) camera mode on iOS and Android devices, follows an active liveness protocol (zoom-in, zoom-out, head turns, natural blinking)
The dataset targets iBeta Level 1 PAD certification - the entry tier of biometric presentation attack detection testing. iBeta Level 1 tests biometric systems against ISO/IEC 30107-3 for 2D attack vectors including paper prints and display replays. This is typically the first certification tier sought by fintech, KYC, and identity verification providers
The Level 1 Dataset covers 2D presentation attacks: paper prints, cutouts, and display replays, which are easier to defend against but far more common in real-world fraud. The Level 2 Dataset covers 3D presentation attacks: silicone, latex, wrapped 3D paper, and cloth masks, which are rarer but harder to detect. Most fintech and KYC providers begin with Level 1 certification before progressing to Level 2
Paper attacks were captured on iOS and Android phones against printed photos, cutouts, cylinder-mounted prints, and 3D paper masks of participants' own selfies, with controlled variation in angle, distance, and lighting. Replay attacks were captured by re-recording participants' own videos on smartphone, laptop, and PC displays across varying brightness, distance, and viewing angles
Yes. 21% of companies that passed iBeta certification in 2025 are Axon Labs clients. See the iBeta Certification Success Stories section for detailed case studies, including a UK fintech that passed iBeta 2 after a failed attempt, 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
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. Participants contributed their own selfies and reference videos as source material, providing clear legal basis for the subsequent attack construction. The dataset is licensed for commercial use in AI model training, validation, and iBeta certification preparation
Yes. A sample version of this dataset is available on request, you can verify attack diversity across both paper and replay sub-collections, video quality, active liveness sequences, and format compatibility with your training pipeline before committing to the full dataset. Submit a request 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
Didn't find what you were looking for?
Our collection includes many datasets for various requests
iBeta Level 1 Dataset
– 35,000+ videos
– 85+ participants
– zoom in and
zoom out
iBeta Level 2 Dataset
– 25 000+ videos
– 3D masks
– iBeta Level 2
iBeta Level 3 Dataset
– 10,000+ videos
– 12 Unique Masks
– iBeta Level 3
Display Replay Dataset for Liveness Detection
– 9,000+ videos
– 6,500+ participants
– Balanced mix of genders and ethnicities
