Paravision white paper series guides PAD and biometrics implementations for health and air travel
Paravision has published a series of three white papers on how to use biometric facial recognition and presentation attack detection to support safe economic reopening in a range of different scenarios, from office health protocol enforcement to remote onboarding.
The white papers on ‘COVID-19: Enhanced Safety & Security,’ ‘Face Recognition and the Future Air Travel Experience,’ and ‘An Introduction to Presentation Attack Detection’ are available free for download.
Paravision Chief Product Officer Joey Pritikin tells Biometric Update in an email that in terms of accuracy, robustness, usability, and price-performance, face recognition has become compelling for a number of use cases biometrics may not have been appropriate for in the past.
“Because of this, we are getting approached by many end users and partners who don’t have a background in biometrics and therefore don’t have insight and context for many of the technologies and product features that are relevant to their use cases,” he explains.
“Our goal with these white papers—which are the first in a broad series—is to help systems integrators and solution providers as well as end customers to build a solid foundation for understanding the key technologies, applications, and deployment considerations associated with face recognition, biometric identity, and computer vision. Regardless of whose technology they use, we want to help business leaders to be thoughtful and informed when considering the role that technology can play for them.”
The use of facial recognition to meet the requirements of the “new normal” safely and securely is examined in the “Safety & Security” white paper, in particular facial recognition based on robust biometric algorithms that can accurately identify faces partially occluded by masks. The use of the technology for detecting masks and other personal protective equipment (PPE), as well as social distancing and activity tracking is also explained.
Paravision outlines workplace face biometrics and computer vision applications for preventing COVID-19 transmission including touchless access control and safety policy adherence. In commercial settings the technology can help reduce contact and adapt customer experiences, while also helping support safe retail practices like sanitation. Paravision’s facial recognition can also be used at airports for seamless travel solutions and policy conformance.
The latter use case is explored in greater detail in Paravision’s white paper on the “Future Air Travel Experience.” This white paper considers the processing and image capturing approaches used in airports, and explains the use of facial recognition for remote registration from the comfort of one’s home or office, check-in and baggage drop processes, security checkpoints, immigration processing, wayfinding, access to VIP lounges and boarding.
Pritikin notes that Paravision’s face recognition was developed to work with occluded faces, so has not been altered to work with masks. The company’s technology was ranked second in accuracy by NIST in a test of pre-COVID algorithms. Paravision did develop a mask detection model, however, using what Pritikin describes as “a highly efficient machine learning infrastructure” to build “a very robust model in a matter of weeks.”
“Analytics for social distancing are a similar consideration,” Pritikin explains. “Person detection is a critical capability for many applications in video security where there is a desire to combine security and free-flowing movement. Social distancing merely puts specific metrics on top of the models for person detection, and so the effort to accomplish it is minimal when based on a solid computer vision toolset.”
The white paper on presentation attack detection (PAD) technology introduces the concept of biometric spoofing, how it works and how it is detected, and the three levels of presentation attacks. Paravision considers PAD technology based on whether active or passive participation is required from the user, as well as standard versus specialized camera technology, with advantages and disadvantages noted for each. Finally, four common use cases for PAD technology are pointed out.
Paravision further upgraded its biometric accuracy with a 36 percent reduction in error rate from the latest release of its face recognition algorithm.