Ayonix’s 3D face recognition technology has been developed by Ayonix R&D engineers where all Ph.D holder scientists are heavily working on state-of-the-art AI-based 3D Face recognition technologies. 3D Face technologies consist of Face detection, Multi Face tracker, 3D Face construction, Local face feature detection and extraction, Face feature grouping.

State-of-the-art FR Technologies

  • ARM-CPU friendly architecture
  • Match under a second for 1 million records
  • 99% accuracy with very small error rate
  • Powerful Deep learning algorithm
  • Highly customizable and flexible structure
  • NIST-tested&approved technology
  • Minimized hardware resource requirement
  • Ultra-compact engine
  • Super fast matching
  • Support for complex environments
  • Multi-people detect

Ayonix scientists work with Software and Product development departments closely. Ayonix scientists are aware of what industry needs, how technology should help people. For example, Ayonix scientists discovered that people wear face mask because of Covid-19 which makes face recognition difficult to recognize. They developed a new face recognition technology to recognize people from masked faces in a 98% accuracy.

Ayonix Scientists in the R&D department are very motivated to provide a human-like face technology which makes Ayonix incomparably advanced company.

From its earliest research 20 years ago to detect and trace eye locations, Ayonix has evolved into one of today’s leading face recognition software developers, building cutting-edge technology using its patented and proprietary algorithms.

The Ayonix platform ranks among the world’s best for speed, accuracy and scalability. Developed completely in-house, our advanced, military-grade AI system applies advanced mathematical approaches and pattern recognition to detect and match faces.

With lightning speed, our algorithm converts 2D video or still images into data, which is then analyzed against statistical norms to reconstruct faces in 3D. Bolstered with an intelligent deep-learning network, we process each image in under 10 milliseconds which is at least 10 times faster than the near competitors from our global competitors.

  • Ayonix uses unique mathematical approaches that convert all non-frontal faces to frontal images, modifying facial elements to standard forms. If mouth is open, we close; if eyes are close, we reconstruct to open; if there is an angle, we interpolate; if light is bad, we improve.
  • An intelligent feature algorithm homes in on faces that are obscured by facial hair and sunglasses or have different skin tones.
  • Ayonix technologies require significantly lower hardware and related technology investments, allowing customers to deploy resources smartly.

Steps of the Face recognition

Capture a facial image and identify key points at the eyes, nose and mouth

Extrapolate 100 points on the face based on the initial image capture – and compounds that data to build thousands of points in between

Apply a 3D mesh, which corrects for variables such as shadows, front hair style and a subject’s facial angle.

Extract gender, age and person-specific features from the face image – akin to an analog to digital conversion.

Match face to databases and generate matching score between 0 to 100(100 is highest score)