Free Single-Frame 2D Liveness Checks

On August 31st, 2020 FaceTec announced Unlimited Free 2D Liveness for any Customer, Partner or Customer of a Partner.  If you are not yet a FaceTec customer, please visit www.Liveness.com and contact a Partner in your geography from the FaceTec Certified Partner list. (Partner Participation in this offer may vary)
 

Introduction

An overview and technical information about the 2D Liveness Check feature in the FaceTec Server. Before using 2D Liveness, be sure to read the Accuracy section below.

The 2D Liveness Check works with camera resolutions as low as .3 megapixel (face should be at least 150px wide) and DOES NOT Require a FaceTec Device SDK to capture the image, allowing for basic anti-spoofing security to be provided when processing a single photo frame captured in a semi-supervised environment such as an Airport Customs checkpoint or a Self-checkout in retail store.

The 2D photo is processed by the 2D Single Frame Image Liveness Check can then be matched to a previously enrolled 3D FaceMap using the 2D Photo:3D FaceMap Matching. This enables a user to enroll remotely and then authenticate locally to their previous 3D FaceMap at up to 1/2,000,000 FAR.

 

  Spoof Example                                                      Real Example

Use Cases & Behavior

The 2D Spoof Check API is designed to run on normal to high quality, 1st Generation (captured from a live person & unaltered) images. Even if the 2D image was captured 1st generation, if the image has filters applied, red-eye removal, or had otherwise been edited it will likely be rejected by the 2D Spoof Check. 

Face pose & lighting conformance to ISO 19794-5 is preferred and deviations from best practices will result in poor/undefined behavior. Images should be frontal, upright, neutral expressions, with no excessive bright or shadows, and the face is >50% the total image area and the face size is >150px wide, from ear to ear.

Supported Images & Use Cases

  1. 2D Portraits like Government Passport/Driver License photos from supervised captures.

  2. In general, original images that are Face Portraits, where the face is >150px wide in the image, the image is good quality, and the subject is looking directly into the camera.

  3. .3mp cameras and above.

Images & Use Cases that are NOT Supported

These use cases are not allowed and not supported:

  1. 2D Images that have had filters applied, been cropped, enlarged, skewed, or otherwise edited.

  2. Deepfake images or video frames.  Please use FaceTec 3D Liveness Check to detect these.

  3. Tiny or low quality images.

  4. No Photos of ID Documents, Driver Licenses or Passports, for those use Photo ID Anti-Spoofing.

  5. Images where environment conditions are particularly bad.  For instance, excessively blurry, lighting that causes data loss on the face, glare in the glasses, partial occlusions on the face, sunglasses, pitch black images or images where otherwise the data on the face is ‘blown out’, excessively off-angle faces, etc.

Accuracy

The 2D Liveness Check API is +98% Accurate on Live Humans and at blocking Level 1,2 & 3 Spoof types with .3mp camera from smart phones, tablets and webcams, security cams and others when tested with random attacks.

To achieve this accuracy, the photo must be a mugshot style Face Portrait, where the face is >150px wide in the image, the image is good quality, and the subject is looking directly into the camera lens.

FaceTec performs basic quality checks on the 2D image prior to rendering a “Likely Real” or “Likely Spoof” decision, but your results may vary if you deviate from the recommended image contents and allowed use cases described above.