Facial Recognition
Incode’s AI doesn’t just compare faces — it examines their biometric structure to distinguish subtle features in different contexts and with unparalleled precision.
Top companies switch to Incode for our proven impact on fraud protection and growth
Threat prevention with advanced facial recognition
The rise of accessible Generative AI (GenAI) has made it easier than ever to commit identity fraud with digital impersonation
GenAI enables fraudsters to impersonate and create fake identities with as little effort as a single photo of the victim. Incode’s facial recognition, trained by millions of verifications monthly, stays ahead of these threats, offering businesses and customers essential protection against sophisticated biometric and deepfake fraud.
Setting the gold standard in facial recognition
Our superior facial recognition is powered by globally inclusive and diverse training data, enabling near errorless performance across ethnicities, age, and gender
Demographic equity
Unbiased performance across demographics
Environmental adaptability
Reliable results in any condition
Accuracy at scale
Exceptionally low occurrence of less than 0.01% false matches
Database performance
Facial matching completes in 20 milliseconds
Interoperability
Across NIST-compliant businesses
Precision in every pixel
Our Facial Recognition technology uses advanced ML models to compare selfies with ID photos or previously captured selfies stored in the database. This multi-pronged approach ensures accurate identity verification and fortifies fraud prevention while maintaining a seamless user experience.
How it works
How it works
Face detection
Identifies and isolates unique facial features from an image or video for subsequent analysis.
Feature extraction
The system analyzes the face to identify and extract unique features – high-dimensional representations of complex facial patterns and unique characteristics, ensuring accurate recognition despite changes in expression or lighting.
Vector conversion
The extracted features are converted into a numeric representation of the facial biometrics. This securely encrypted data vector is a unique “facial signature” that captures the essence of the face.
Encryption for security
The vector containing this facial signature is then securely encrypted into a format that can only be opened and interpreted with the correct decryption key, thereby safeguarding against unauthorized access and breaches.
Comparison for verification
The facial recognition verifies by comparing selfie captures and ID images after their conversion and encryption as vectors. These encrypted vectors are matched in the applicable of two possible scenarios: Selfie vs. ID (1:1), verifying the capture against the ID capture, and selfie vs. database (1:N), verifying the capture against captures stored in Incode’s database, to generate a result.
Certified by the National Institute of Standards and Technology (NIST), Incode’s Facial Recognition ML models meet rigorous international standards for accuracy and security – leading the pack when testing millions of images for fraud
- Ranked among the top performers in NIST FRTE benchmarks, including 1:1 verification and 1:N identification
- 100% success rate in identifying and blocking fraudulent selfies
- 99.9% success rate in identifying and passing genuine selfies
- Facial Recognition ML models process verifications and deliver results in 20 milliseconds
“Incode’s SDK integrated seamlessly into our systems. Its user experience led to a stronger completion rate throughout our onboarding process.”
Product Leader
at Nubank
