top of page
Facial Recognition.jpeg

FACIAL RECOGNITION

Biometric Facial Recognition

United Smart Tracker is a state-of-the-art biometric solution designed for real-time face detection and recognition. Deep learning neural networks algorithms underlying in the principles of United Smart Tracker operation boost security at crowded areas.

United Smart Tracker performs the facial recognition process by comparing facial features from given face image with face images within a database. Face images can be previously uploaded to the database from any source: photographs, video recordings, scanners, web cameras, photo cameras, CCTV cameras, etc. Biometric search can be performed both in real time and on request.

United Smart Tracker face recognition algorithms adapt to lighting conditions and do not depend on appearance modifications related to aging, beard, hairstyle changes, etc.

United Smart Tracker can be used for a wide range of potential applications:

Airports and other transport hubs
Detect and track suspects in real time to immediately respond to potential threats and prevent panic

City infrastructure
Search individuals through database of wanted persons to create a safer environment

Stadiums and other sports venues
Prevent blacklisted individuals from entering sporting events or concerts

Key Features and Advantages

  • Face detection in live stream

  • Biometric search in real time and on request

  • Flexible, easy to deploy, no extensive IT support required to implement

  • Any face image source can be used for further facial recognition

  • Face image quality evaluation according to ISO/IEC standard 19794-5

  • Password-protected access and user rights administration

  • User activity monitoring and logging

  • Immediate alert notifications sending to operators’ desktops, by e-mail, and SMS

  • Multi-user web-based interface

  • HTTPS protocol to link United Smart Tracker components

Technical Characteristics

Accuracy

97% of correct matches in case of 1% of false acceptance rate / 100% correctly matched facial samples in 1 million database in case of analysis by the operator 10 best results on the database according to ISO/IEC standard 19794-5

Performance

  • Identification result in 1-3 seconds

  • Biometric search in 1 million database in 3 seconds

  • Minimal image requirements for facial recognition

  • Pose up to +/-15˚ deviation from frontal image

  • Batch input and integration with the existing databases

Failure tolerance

  • Fully distributed architecture

  • Back-up at all levels and redundancy

  • Quick recovery and robustness to hardware malfunction

bottom of page