|
Recent analysis of facial recognition systems
has indicated that greater robustness can be achieved with
better understanding of the image and scene in the image provided
to a facial recognition system. This relates specifically
to understanding and coping with variation in lighting, and
having implicit knowledge of the distance, position and angle
of a subject's face.
faceLAB™ has been developed over
a number of years in parallel to the Biometrics industry,
for quite different reasons, but always with these elements
in mind: robustness, tolerance to lighting variation, and
establishing reliable, constantly tracking measurements relating
to the face. With technology based on a fully licensable core,
faceLAB™ provide automatic tracking—not just
face finding—complete with implicit knowledge of scale,
rotation and distance in scene.
 With
automatic acquisition, a wide field of view, and constant
evaluation of the scene in view, faceLAB™ technology
is uniquely situated to prototype improvements to existing
facial recognition.
faceLAB™ produces accurate data on-time, in the
toughest of tracking environments. From outdoor intelligent
vehicle or aerospace studies through to screen & web usability,
stimulus response and laboratory environments, faceLAB™
provides high accuracy and tolerance to a wide range of operational
conditions.
New in faceLAB™ v4 is a leading-edge active illumination
technique, increasing the fidelity and accuracy of gaze assessment.
Track each eye independently, and if the subject is close
enough, even track pupil
size and vergence.
If conditions degrade, faceLAB™ continues to maintain
the confidence of its classic, robust tracking system by default.
Analyse and model fatigue and drowsiness in real-time with
faceLAB™'s true blink analysis and PERCLOS fatigue
assessment, including data on the minutia of eyelid behavior.
Measurements are taken accurately on eyelid position rather
than bright pupil or corneal occlusion.
Eliminate the hassle of interference, setup and correlation
issues associated with traditional contact-based tracking
methods—faceLAB™ does not use any RF, ultrasonic
or electromagnetic systems. Our unique visual tracking system
provides a single stream of data. No need to stitch or align
multiple tracking devices to obtain a result. No painstaking
calibration of separate head tracking systems—faceLAB™
itself caters to changes in pose or the movement of a subject.
Specific techniques have been implemented in faceLAB™
v4 to enable easier calibration of children and uncooperative
subjects.
Minimise the time, cost and headache associated with experiments
with faceLAB™'s unparalleled ease of use, quality
of support and user-friendly interface. faceLAB™'s
hallmark automatic initialisation feature has been improved,
and now produces raw gaze output automatically, as well as
head position and blink events over a wider
field of view. Laboratory or simulator implementations
are further enhanced—a wide variety of digital simulation
platforms and third-party analysis
tools are now natively supported. Get direct representation
of subject interaction inside your simulation environment,
in real-time, and synchronised correctly to the experiment.
Don't restrict your eye tracking experiments to screens—faceLAB™
continues to offer a fully integrated 3D world model that
encompasses physical objects as well as displays.
Maintain data and tracking integrity through a range of lighting
and movement conditions. faceLAB™ immediately
recovers if a subject leaves the field of view, and the system
is robust to occlusion or sudden movement. Work with all faces
and all eye types, with or without contact lenses and with
most eye-glasses. Work in bright sunlight or at night time,
with subjects close to the camera or several metres away.
Even if you can't see the eyes—if the subject is wearing
sunglasses for instance—faceLAB™ still finds
the head pose, eyes and mouth, and continues to produce accurate,
robust and reliable data. Added flexibility is provided with
the new ultra-wide
field of view mode, allowing experiments with distant
subjects or a variable experimental area. Reduce the amount
of data stored or transmitted with variable width logging—capture
only the data you need, and send it across a network where
you need it.
Develop real-time applications using the faceLAB™
freely available SDK, enabling other systems to localise salient
facial features such as the eyes and mouth, in real-time,
across standard network interfaces.
Seeing Machines has simplified the path to commercialising
products based on faceLAB™ technology with a licensable
core now available.
Contact biometrics@seeingmachines.com
if you are interested in specific development or licensing.
|
