Human beings are really good at recognizing faces, it is an innate ability, evolved over deep time. We are now applying this natural trait within our technology stack to add the digital equivalent of this capability, facial recognition technology, to many applications.
Being able to recognize faces can be thought of as a sort of human ‘super-power’ - it allows us to pick out a face in a thousand. Now, facial recognition technology is being applied to a variety of industries to do a similar thing. These applications are pushing the market for facial recognition technology to grow at a rate of around 17% CAGR with a value predicted to be worth $9.78 billion by 2023. The technology is being applied to many areas such as access control, criminal detection, tailored marketing campaigns, and health assessments. Facial Recognition works by using data, in this case, image markers, and matching them to existing database entries. In much the same way that fingerprints are mapped to identify unique features, faces can also be mapped. This ‘faceprint’ is analyzed and cross-referenced against existing ‘faceprints’. New generation facial recognition technologies are powered using deep learning, a subset of Artificial Intelligence that provides a highly accurate analysis.
What is Driving Facial Recognition Technology?
A number of market analysts have identified drivers for the rapid uptake of facial recognition systems, these include crime prevention - the increase in the use of the technologies for surveillance, border control, and general crime prevention. In addition, the use of facial analytics and deep learning are improving match rates and outcomes of this application of the technology. Increasing identity theft - Identity-related fraud is increasing year on year.
The wide-ranging applications of facial recognition are moving the technology into more commercial use cases. Typical IP cameras in retail outlets, fitted with facial recognition software, are being used to create smarter marketing and theft prevention by gathering intelligence on shoppers such as their predicted age, gender, dwell time etc.
What are the Security Concerns with Facial Recognition?
Our face is a very intimate part of our being and certain concerns have been raised in terms of facial recognition technology. Amongst the various ethical concerns, certain security ones stand out, examples include:
- False matches - Facial recognition has a perfect use for crime or theft prevention. However, the false alarms of false positives is worrying, as the result of being incorrectly identified by software, could result in a wrongful conviction. However, this is being addressed as the technology matures. Intelligent face matching using deep learning facial recognition are improving match rates significantly.
- Authentication - Controversy has surrounded certain applications of facial recognition used to authenticate users to, for example, mobile devices. Apple’s iPhoneX which uses the technology was recently lambasted for being ‘racist’. For example, the colleague of a Chinese woman was able to unlock her phone using her own face - the woman accused Apple of using ‘white faces only’ to train the software.
Both of the above issues are being addressed as the technology becomes more intelligent. The use of deep learning-based algorithms, are improving the quality and ultimately the security of facial recognition. In addition, the National Institute of Standards and Technology (NIST) are working with industry partners to develop standards. They state, “NIST is actively pursuing the standards and measurement research necessary to deploy interoperable, secure, reliable and usable identity management systems.”
What’s Happening Across the World of Facial Recognition
The security world is exploring facial recognition at a fast pace. Projects are growing across the world and if you aren’t doing it now it is likely to be on your near-term roadmap. Here are some examples of where the technology is being used in real-life:
Preventing Retail Crime:
One of the natural fit areas with facial recognition capability is the retail sector. Retail crime is a major issue across the world. The amount of financial loses in the industry, in 2017, was $46.8 billion, with shoplifting and organized retail crime being the two most prevalent issues. AI Facial Recognition is able to help prevent shoplifting. The system works by referencing a database of known offenders with shoppers images. These are then used to prevent future crimes. In an LPM survey, a full 80% of shoplifters said they were habitual shoplifters because they knew they could get away with it. Having an intelligent system as a strong deterrent can help to reduce repeat crime. The AxxonSoft system has shown both a successful return on investment in retail stores with 3-4 criminals being caught per week per store among 10 000 visitors per week. Though the system is not ideal and has 10 false alarms per camera per day, it is fully satisfying the customer.
