 |
Overview | What is faceAPI | FAQ | Specs | License Enquiries
|
The Ultimate Real-Time Face Tracking Toolkit for Developers and OEM's
|
What is the faceAPI?

- API stands for Application Programming Interface.
- “faceAPI” is short for “Seeing Machines Face Tracking API”, which contains a set of binaries, developer tools, sample code and documentation. The core component is a windows shared library (DLL) and a set of C header files.
- The faceAPI is a developer toolkit for embedding head-tracking algorithms into any application on Windows XP.
- As of version 2.1, the API consists of 83 functions.
- Multiple head-trackers can be run on the same system at the same time.
- These functions allow a programmer to create and control two types of head-tracker, using any WDM compatible camera installed on the system. In a nutshell, a program that uses the API would proceed by calling API functions to:
- Return a list of cameras available on the system.
- Create a head-tracker of a specified type, using a particular camera. The API then initializes and controls the camera.
- Register a callback routine to receive minimum latency tracking data from the head-tracker.
- Create any number of video display windows to show live video and tracking information. These can be embedded in the programmer’s application.
- Control the tracker – e.g. starting tracking, controlling how the tracker behaves when the tracking is lost, controlling what is shown on the video window.

Each head-tracker produces “head-pose” data in 3D world-coordinates, at the frame-rate of the camera. This data is passed a simple structure to the user-defined callback routine. Head-pose data consists of:
1. 3D Head position in space along X,Y,Z axes. The origin of the head is defined as the midpoint between the eyeball centres.
2. 3D Head rotation in space around X,Y,Z axes.
3. 3D Position of left and right eyeballs in space.
|
 |
|
faceAPI features
- Highly robust, real-time, monocular face tracking;
- 6 degree-of-freedom (3D) head tracking: (X, Y, Z position and rotation);
- Automatic initialization – automatic face finding and commencement of tracking, with automatic reacquisition if tracking is lost;
- Facial feature detection – extracts and outputs facial feature coordinates;
- Can track up to +/- 90 degrees of head rotation;
- Can track with as few as 50 pixels across the face;
- Large tracking volume (as a guide, a 640x480 camera provides tracking volume of approximately 100 x 60 x 100 cm)
- Works with any Windows-based web camera;
- Designed for all human faces: robust to occlusions, fast movements, large head rotations, lighting, facial deformation, skin color, beards, glasses etc;
- Full control over all tracking parameters for purposes of software integration.
|



|
|