Skip to content

Overview#

Luna SDK contains two primary libraries:

  • FaceEngine SDK is the core library. It is described in the FaceEngine_Handbook.pdf;
  • LivenessEngine SDK includes complex liveness checks. It is described in the LivenessEngine_Handbook.pdf.

The TrackEngine tool is used for receiving images from video.

See Licensing.pdf for information about features licensing and license activation.

See FeatureMap.htm for additional information about available functionality and features.

The ConfigurationGuide.pdf includes information about SDK configurations.

FaceEngine#

FaceEngine SDK enables you to implement the following functionality:

  • Faces and human body detection in still images;
  • Landmarks location on faces and human bodies;
  • Faces warping;
  • Human body resize;
  • Face and human descriptor extraction and matching;
  • Descriptor search index creation (fast Hierarchical Navigable World graph or HNSW shortly);
  • Batching for processing performance improvements.

FaceEngineSDK library is available in two versions: Frontend Edition (FE) and Complete Edition (CE):

  • Complete Edition supports all functionality;
  • Frontend Edition is a light version with excluded descriptor functionality (does not support descriptor extraction and matching).

LivenessEngineSDK library is designed to distinguish a real user’s face from a fake one (e.g., printed on paper or displayed on the screen). LivenessEngine uses FaceEngineSDK library modules and performs detection, warping, and various estimations.

TrackEngine#

It is a tool for face detection and tracking on multiple sources. It allows to pick the most suitable still images for facial recognition from a sequence of video frames.

Note, that TrackEngine itself does not perform any facial recognition. It’s purpose is to prepare required data for external systems.