What is the difference between Face Comparing (1:1) and Face Searching (1:N)? 
Face Comparing verifies whether two faces belong to the same person—ideal for identity authentication scenarios like login verification or ID document matching. Face Searching identifies a face from a large database—used for finding a person among thousands of enrolled faces.
What use cases is Face Comparing designed for?
Common applications include: account login verification, remote identity proofing, payment authentication, access control systems, and border/ID document checks where you need to confirm "the person presenting the credential is the credential owner."
What input does the API require?
Two face images or face tokens. Images can be uploaded as Base64-encoded strings, image URLs, or face tokens from previous detections.
What does the confidence score represent?
The confidence score indicates the probability that the two faces belong to the same person, typically ranging from 0 to 100. Higher scores indicate greater similarity.
What are thresholds and how should I set them?
Thresholds are configurable cutoffs that determine match/no-match decisions. We recommend:
- High-security scenarios (financial transactions, border control): 80+
- Standard authentication (app login, employee access): 70-80
- Low-friction experiences (content personalization): 60-70
Does Face Comparing work across age, pose, and lighting variations?
Yes. Our algorithms are trained on diverse global datasets and maintain high accuracy across age ranges, facial expressions, partial occlusions (glasses, masks), and varying illumination conditions.
Is face data stored or retained after comparison?
By default, no biometric data is retained. Face tokens can be configured with custom expiration times, and all processing complies with GDPR, CCPA, and ISO 27001 standards.
Can Face Comparing run on edge devices?
Yes. Face++ provides SDKs for mobile (iOS/Android) and embedded systems, enabling offline 1:1 verification without cloud dependency.
