SAFR SDK Documentation

The SAFR SDK is a set of shared libraries that collectively implement an object tracking mechanism, face recognition, and event reporting. The object tracker is used to locate and track different types of objects (e.g. human faces or badges) in a video file or live video stream. Face recognition can learn new identities or match existing identities. The event reporter notifies your application of detection and recognition events. Object detection, face recognition, and event reporting are performed in real-time.

SDK are supported for the following OS's:

SAFR Embedded SDK

Embedded SDK should be used when the facial recognition software needs to run on a device with limited resources (e.g. RAM, CPU, or memory) without any internet or cloud connectivity. However, the device must still be able to perform tasks such as detection and recognition of images with varying conditions (e.g. insufficient lighting, people wearing hats or glasses, etc.) with a minimum response time.

Embedded SDK are supported for the following OS's:


SAFR exposes REST API which can be called remotely or locally to access a wide variety of SAFR services. An overview of the REST API is located here:

OpenAPI reference documentation of the REST API can be found at the following locations:

OpenAPI Docs URL Description
SAFR Computer Vision API (COVI) Main service API used to enable face detection, recognition, and management of recognized persons and faces.
SAFR Computer Vision Events Server API (CVEV) Used to publish, query, modify, and delete recognition events.
SAFR VIRGA Server API (VIRGA) Used to configure video feeds for monitoring object recognition.
SAFR Object Server API (CVOS) Used to manage object storage and data stream access.