The Ultimate Real-Time Face Tracking Toolkit for Developers and OEM's

FAQ

1. Is there a free version available?
2. What information does the tracker provide?
3. Anything else?
4. What is the minimum system configuration?
5. Does it run on Vista or any other OS?
6. Can I track multiple people at the same time?
7. What languages do you support?
8. Will it work on anyone?
9. What if I’m wearing sunglasses or thick spectacles?
10. Does it work in dark conditions or bright sunlight?
11. Can you track other things besides faces?
12. Where did this technology come from?
13. What are some applications for this technology?

1. Is there a free version available?

• Not yet. This may change in future – please keep an eye on the website.
  

2. What information does the tracker provide?

• Head position (X,Y,Z) relative to the camera, in meters.
  
• Head rotation (X,Y,Z) relative to the camera, in radians.

3. Anything else?

• No – just 3D head position & rotation tracking at the moment.
  

4. What is the minimum system configuration?

• Windows XP SP2
• Core-2 Duo
• A good quality USB 2.0 webcam (recommend Logitech Quickcam Pro or similar)

5. Does it run on Vista or any other OS?

• Not yet. We plan to release the API under all major operating systems including Linux and Mac.

6. Can I track multiple people at the same time?

• Not yet. If there are multiple people in the image, the tracker locks onto the “best face”.

7. What languages do you support?

• Currently, the API is a low-level C library.
• We plan to release a set of open-source ANSI C++ wrappers, and managed C++ version for easier inclusion into .NET projects.
• A Python wrapper is also on the roadmap.

8. Will it work on anyone?

• Yes we think it does. We obviously can’t prove this, but only assure you that we have a database of many thousands of faces.

9. What if I’m wearing sunglasses or thick spectacles?

• It still works.

10. Does it work in dark conditions or bright sunlight?

• Yes. The main issue is frame-rate. Webcams often throttle to low frame-rates in dark conditions. Tracking at 15Hz works, but not nearly as well as 20 or 30Hz.
• If the image is completely black or saturated white, then obviously there is nothing the camera can “see” to track, however, for example, half the face can be covered and it still tracks fairly well.

11. Can you track other things besides faces?

• No the faceAPI is designed for human faces. We don’t do hands, bodies, dogs, monkeys, cats, chickens or {insert strange thing here}.
• We are focussed on building the world’s most automatic and robust face-tracking algorithms for advanced HCI applications.

12. Where did this technology come from?

• This technology has been in development for over 8 years inside Seeing Machines Ltd.
• Original versions ran using stereo-camera rigs and it was originally designed for enhancing robot-human interaction.
• It then became part of faceLAB, a research tool for studying human behaviour under real-world conditions.
• The first large-scale application for the technology is the detection of drowsy and inattentive drivers, which is why the tracking has to be incredibly robust and completely automatic. See our DSS product.
• We are now opening up this technology for other people (you) to develop applications that leverage face-tracking.

13. What are some applications for this technology?

• There are so many ideas, we can’t keep track of them all. It’s up to you, which is why we have built this API. Go forth and innovate!