Datasets:
metadata
license: cc-by-nc-4.0
task_categories:
- image-feature-extraction
- image-classification
- video-classification
language:
- en
tags:
- liveness detection
- anti-spoofing
- biometrics
- facial recognition
- machine learning
- deep learning
- AI
- paper mask attack
- iBeta certification
- PAD attack
- security
- ibeta
- face recognition
- pad
- authentication
- fraud
7,000+ people, 70,000+ images: current selfies + archive photos
Contact us and share your feedback - recieve additional samples for free! 😊
Real Multi-Year Temporal Coverage
- Archive photos from 6 months to 10+ years before current images - not synthetic aging
- Natural aging progression: wrinkles, hairstyle changes, weight fluctuations
Above-Average Photos Per Person
- 8-15 archive photos (past) - historical photos from 1-15 years ago
- 2-4 current selfies (present) - recent appearance with lighting and environment variations
Total: 70,000+ images across 7,000+ identities
Diverse Real-World Conditions
- Multi-device captures: Android, iOS, Windows (phones, webcams)
- Multi-ethnic coverage: Balanced representation across demographics
- Age range: 18-70 years old
Metadata Included
- Demographics: gender, ethnicity, age
- Device info: OS (Android/iOS/Windows), device type, model
- Temporal data: archive photo years
Full version of dataset is availible for commercial usage - leave a request on our website Axonlabs to purchase the dataset 💰
Potential Use Cases
- Face Recognition & Verification - High sample density per person for robust training
- Temporal/Age-Invariant Recognition - Multi-year gaps for aging-robust models
- Demographic Bias Testing - Balanced ethnic and age distribution
- Cross-Device Training - Photos from multiple device types and resolutions
- Long-term Identity Verification - Systems requiring recognition across years
Related Datasets with Additional Features
This dataset focuses on temporal face recognition with current and archive selfies. For additional use cases, check out our related datasets:
1. Selfie & Official ID Photo Dataset
- Current and historical selfies PLUS official ID document photos (2 per person)
- Ideal for: KYC verification, selfie-to-ID matching, identity verification
- 6,000+ people with selfies + ID photos + archive photos
- Use case: Training models to match casual selfies against official documents
2. Face Recognition Dataset: Selfies & Videos
- Same base dataset PLUS NIST liveness videos (2 per person: zoom-in + head turn)
- Ideal for: Liveness detection, anti-spoofing, video face recognition
- 1,000+ people with selfies + archive photos + liveness videos
- Use case: Training models to detect presentation attacks and verify live persons
keywords: face recognition, face detection, biometric authentication, liveness detection, anti-spoofing, video face recognition, computer vision, selfie dataset, video dataset