Face Processing

Human Face Processing
Detection, tracking, extraction, and analysis for many applications

Since 2002, Quantum Signal has been developing a very high performance, robust set of algorithms for the detection, tracking, extraction, and analysis of human faces. These algorithms grew from a multi-year project conducted by the company for the National Institute of Standards and Technology (NIST), and go far beyond the simple color-filter or motion detection schemes being used by others. The set of technologies can mine face images in a wide variety of environments, even with poor imagery, and have been tested extensively. The algorithms are compatible with all sources of video and camera types, and are robust to common compression types. And almost all of the algorithms run in real-time!

facial detection images
images of facial expresssions

QS face mining technologies include:

  • Faster-than-real-time detection and extraction of faces in still images
  • Real-time, simultaneous detection and tracking of multiple faces in video feeds
  • Facial feature detection, location, and masking
  • Pose Estimation
  • Detection of chin-lines and hair-face interface and exclusion of non-face regions

QS has incorporated such technologies into key customer products, including FACES and Hunter SmartShot®. Some of the applications of QS face technology include:

  • Building a facial "log book" of visitors to a facility
  • Mining documents, websites, or television for faces
  • Scanning crowds for persons of interest
  • Video games using face or head motion as inputs
  • Occupant sensing for automotive vehicles

Quantum Signal is available to work with you to adapt and incorporate these technologies into your product, or conceptualize and develop entirely new products based on them. Quantum also offers consulting and software engineering services in this area, and can assist you in evaluating and testing face-related technologies and implementing a highly functional solution.


Contact Quantum Signalvideo@quantumsignal.com