Are you a developer who is excited about the opportunity of accessing the DeepAffex™ engine but unsure how to begin this process? Outlined here you will find a brief step-by-step description to assist you to immediately get started on this rewarding journey!
A great place to start is by reviewing our 'DFX Developer‘s Guide' to orient yourself with the features and capabilities of our entire pipeline.
DFX "Software Development Kit" (SDK) Overview
The primary purpose of the SDK is to convert an incoming stream of face-tracked image data into resultant blood-flow. This procedure is referred to as the blood-flow extraction and is an important stage in the TOI “front-end” pipeline functionality required for DeepAffex™ processing. Through configuration, the SDK is used to generate measurement data (binary payloads) sequences that are then forwarded to DeepAffex™ for analysis and processing.
As noted elsewhere the SDK provides extraction of the subjects facial blood-flow information from the incoming video frames. The client application is responsible for supplying video frames with accurate timestamps (e.g. from a digital camera or video file), detecting faces in the video frames and annotating them with "MPEG-4 facial animation points" (a standard naming convention for face point labeling provided by many 3rd party off-the-shelf face-detection libraries e.g. dlib, Visage etc.).
Check back later for updates
The DeepAffex™ SDK exists as proprietary source-code and is written in C. NuraLogix distributes SDK binaries that are compatible with Ubuntu, macOS and Windows operating platforms. We also provide the source files for a C++ wrapper in the SDK. All client applications using the C++ binding will require linking against OpenCV cv::Mat structures. The underlying C library headers have no external dependencies.