Ask or search…
K

Digital Identity Verification

Kairos simplifies user verification by automatically capturing data from the ID document and comparing the ID photo to a selfie provided by the user. The process includes checking the ID for any signs of fraud and verifying the selfie has been taken live, ensuring a secure and straightforward onboarding.

Document Verification

Data Extraction

Kairos uses two approaches to automatically extract data from user ID documents: optical character recognition to gather human-readable data and a read of the machine-readable zone (when present).

Optical Character Recognition (OCR)

Optical character recognition (OCR) is used to scan the document and convert characters into machine-encoded text. This is used to capture the user's name, date of birth, document number, and other personal details on the ID document.
Machine Readable Zone (MRZ):
For documents like passports, MRZ lines (two or three lines at the bottom of the identity page) are a formatted set of characters that contain data in machine-readable format. These lines contain critical information such as the document type, country of issuance, and the holder's details in a standardized format. Kairos extracts data from the MRZ and compares it to the data extracted from the OCR scan.

ID Types:

The most commonly accepted ID types are listed below but check the Supported ID Documents page for a full list:
  • Passports: Preferred for international verification due to standardized formats.
  • Driver's Licenses: Widely used, but formats vary significantly by region.
  • National ID Cards: Common in many countries but vary in format and available data.

Fraud Checks

Kairos checks the document for signs of forgery or tampering. It uses AI models to analyze text and images, detecting any alterations or counterfeit features. This step makes sure that the documents presented are authentic and have not been manipulated. More detail on these checks is provided on the ID Document Verification page in the Product Details section.
Image Forgery: Assesses the likelihood of forgery in the document's visual elements, such as images and backgrounds.
Text Forgery: Analyzes each text character on the document to determine the likelihood of text forgery.
Recapture: Evaluates the probability that the document image was captured off a screen, focusing on pixelation and other recapture indicators.
Screen Detection: Detects the physical presence of display devices like monitors or phones in the document image.
Low Text Confidence: This check flags instances where the OCR-generated text confidence level for key data fields falls below a certain standard.
Artificial Image: Determines the likelihood that the document image was created using photo editing software rather than being a genuine, camera-captured document.
Artificial Text: Assesses whether the texts within the document are likely created with photo editing software rather than originating from a genuine, camera-captured document.

Biometric Verification

Kairos uses a specific face-to-document-face matching technique, designed to more accurately compare selfies with ID document photos than a generic facial recognition algorithm. We also include an effective liveness check algorithm that protects against the most modern spoofing tactics. This approach provides reliable and secure biometric verification for the onboarding process.

Face to Document-Face Matching

Image Capture

The user captures a selfie using a smartphone or webcam. Simultaneously, the user submits a photo of their official ID. The quality of these images is crucial for accurate verification since blur and glare can interfere with accuracy. We have a set of recommended best practices in this section to explain how to get the best results.

Facial Recognition Technology

The software uses advanced facial recognition algorithms to analyze the facial features in both the selfie and the document's photo. It looks for biometric markers such as the distance between the eyes, nose shape, jawline contours, and other facial characteristics. The system then verifies if the two images belong to the same person. This technology is designed to handle common challenges, such as variations in facial expressions, age differences between the two photos, and minor quality issues in the images.

Liveness Detection

Kairos' biometric verification system incorporates a comprehensive range of checks to ensure the authenticity and liveness of the individual being verified. These checks are integral to providing a secure and reliable verification process, safeguarding against a wide array of fraudulent attempts.