What to know when selecting a facial recognition solution

Facial recognition technology has gained traction both in the public and private sector lately, due to its ability to accurately verify and identify people. But, as with several technological developments in recent years, hype has occasionally overtaken reality.  

Chad Parris, President of SRMC, which works with customers in security assessment, master planning and technology design, suggests that there are several factors to consider when employing a facial recognition solution. He also thinks there is a tendency among end users to just install solutions and with little regard for maintenance.

Related article: "Body recognition" added to facial recognition for enhanced security 

“Many customers have high expectations of facial recognition and believe in a ‘set it and forget it’ mentality,” Parris said. “In reality, these systems need a level of care and feeding to assure the highest level of effectiveness. For instance, there needs to be a database of faces to work from to alert the operator of a person of interest. Where this database come from needs to be determined. Will the customer build this or does it come from a third party? For instance, in Las Vegas, casinos share pictures that are used to build the facial database.”
Such a collaborative effort is seen in some other industries as well. Jewelry merchants in India, for instance, have industry associations that maintain databases for the sake of its members. While this is one aspect to be taken into consideration, there are several other factors that customers should know right from the stage where they select a facial recognition solution.

How to select a facial recognition system

The growing popularity of facial recognition technology has given rise to a number of companies bringing out solutions that claim to be the best in the market. Some of the obvious criteria that systems integrators (SI) look for are the number of facial points used to recognize a face, the machine language (ML) framework, cost, and any known challenges that a system would have.

The nature of the site also matters as that will determine several factors including the need for on-premise processing of the captured data.
But there are some things to be considered even after these. According to Parris, customers should do a systematic study of solution providers and their experience.

“The industry has seen facial recognition and other video analytical platforms come and go over the past several years, creating doubts among integrators and consultants,” Parris said. “When considering a facial recognition solution, spend time conducting your due diligence by reviewing the company’s case studies, reference sites of actual installations, integrations with other third-party applications (such as video management systems and cameras), total capital and operating costs and server computer power.”

What to know during installation

After selecting a solution that best suits customer’s requirements, SI should also consider the existing systems in order to ensure its effectiveness. For instance, Parris pointed out that in some cases, using your existing cameras that may be mounted high for general surveillance may not be ideal for facial recognition that needs a more straight-on view.

“Facial recognition software relies on a certain number of pixels on target, angle, lighting, and other factors,” Parris said. “Facial recognition companies will assist with running your existing video through their analytic software to gauge effeteness of exiting camera and determine if the camera needs to be adjusted.”

Beyond the technological aspects, there are regulatory factors that need to be considered as well when installing a facial recognition system. Privacy laws are becoming more strict across the globe and biometric solutions capture sensitive data that warrants robust protection systems.