Facial Recognition
Didit's Facial Recognition solution provides enterprise-grade biometric verification through advanced computer vision and machine learning algorithms. Our system achieves 99.9% accuracy with a false acceptance rate (FAR) of less than 0.1%.
Liveness Detection Methods
Our platform implements three distinct anti-spoofing technologies:
1. 3D Action & Flash
- Employs multi-factor biometric verification
- Randomized action sequence prevents replay attacks
- Dynamic light pattern analysis validates 3D facial topology
- Advanced deep learning models analyze micro-expressions
- Typical processing time: 1-2 seconds
- Recommended for high-security applications (banking, healthcare)
2. 3D Flash
- Utilizes structured light patterns for depth mapping
- Processes 30+ frames per second during capture
- Validates facial topology without user interaction
- Typical processing time: 1-2 seconds
- Ideal for medium-security applications
3. Passive Liveness
- Implements single-frame deep learning analysis
- Detects image artifacts and texture patterns
- Validates facial features using CNN architecture
- Typical processing time: < 1 second
- Suitable for low-friction scenarios
Each method generates a normalized liveness score (0-100%) based on our proprietary scoring algorithm that considers multiple security factors.
Face Matching
Our system performs detailed facial comparison between the best image extracted from the liveness detection and the ID document photo, generating a similarity score from 0% to 100%.
- Processes 68 facial landmarks for precise alignment
- Handles pose variations up to ±45 degrees
- Compensates for lighting variations and aging
- Maintains accuracy across different ethnicities
Configurable Thresholds
You can customize security levels by setting different thresholds for both liveness and face matching scores. For example:
| Score Range | Action |
|---|---|
| < 16% | Decline |
| 16% - 70% | Manual Review |
| > 70% | Approve |
These thresholds can be adjusted based on your risk tolerance and security requirements.
How It Works
Video Selfie Capture
- System guides users through the selected liveness detection method
- Multiple checks ensure optimal video and image quality
Liveness Detection
- Applies the chosen liveness detection method
- Generates a liveness score
- Prevents spoofing attempts using photos, videos, or masks
Face Matching
- Compares the best image extracted from the liveness detection and the ID document photo
- Generates a similarity score
- Applies configured thresholds for automated decisions
Result Processing
- System applies your configured thresholds
- Automatically approves, declines, or flags for review
- Results available via API, dashboard, or webhooks
Key Features
1. Face Detection
- Accurately locate and isolate faces within images
- Works with various angles and lighting conditions
2. Multi-Method Liveness Detection
- Choose from three distinct liveness detection methods
- Configurable security levels
- Liveness scoring from 0-100%
3. Face Recognition
- Advanced facial comparison algorithms
- Similarity scoring from 0-100%
- Configurable matching thresholds
4. Quality Assurance
- Real-time feedback during capture
- Multiple quality checks
- Optimal positioning guidance
For more information about potential warnings during the facial recognition process, check our Facial Recognition Warnings guide.