Dennis Goldenson, Director of Artificial Intelligence and Machine at SAR Insight & Consulting, looks at the range of expanding uses for facial recognition and the importance of regulatory controls.
The other day, as I was walking briskly through a stressful airport, I came across Clear, a biometric authentication technology and kiosk that identifies a passenger’s identity through fingerprint or eye scan. It essentially helps those identified passengers bypass longer airport security lines. As I quickly walked by the kiosk, I stopped for a moment and questioned whether I was ready to give up my privacy but recognised the new reality that eventually my face will become my new ID and password.
Airlines have started using other forms of facial recognition to assist passengers with check bags, check into their flights and board airplanes faster. The advantage of course is to try and make flying less stressful by providing convenience and safety. As the increased wrinkle lines in my face will tell you, anything to make the flying experience easier and dependable is a plus in my book.
The technology behind facial recognition uses a software application to generate a template of a person’s facial image. The face is measured by calculating key nodal point landmarks: the distance between eyes, the width of the nose and length of the jaw for example. The application compares the template to pictures of pre-existing images of a face to detect or verify a person’s identity. The pre-existing photographs can be found in a number of places including driver’s license databases, government identification records, and/or social media sites.
A number of established tech firms such as Amazon, Google, Apple, Microsoft, IBM and Facebook, and smaller start-ups such as SenseTime, Trueface, Megvii, OrCam Technologies, Clarifai, FaceX, Sighthound, and Facefirst, are all developing facial recognition algorithms that are being trained to identify your face. As these algorithms continue to learn, they become more accurate. To improve the recognition rates, emerging technologies such as 3D facial recognition (with 3D sensors on a CMOS chip can capture facial imagery) and biometrics are now deploying unique skin texture detection for more reliable results. Accuracy continues to improve as multi-sensory modes such as touch, voice and vision are considered to enhance authentication abilities and precision. Companies such as Nuance and Sensory are heavily invested in voice and vision/facial fusion capabilities to deliver more accuracy in authentication security.
IC chips powering facial recognition systems require a tailored solution. There are various companies now providing these advanced processing-based solutions, for example: Ambarella, a computer vision IC supplier has developed a new camera SoC for real-time facial recognition on edge devices; and a Chinese company Yitu has launched an AI chip developed for computers to enhance facial recognition. To meet the complex processing demands of facial recognition, IC chip manufacturers are now developing customised AI chips capable of running a number of applications to support the growth of computer vision.
Use Case Applications
The emergence and capabilities of facial recognition has led to a wide proliferation of this technology. As a result, there are a number of use case applications. Some of these include the ability to unlock and access control on the smartphone (i.e. iPhone X), protect law enforcement and prevent crime, improve targeted marketing and advertising, secure financial transactions, and make air travel safer and more convenient. Facial recognition is being deployed in various locations including city streets, retail stores, restaurants, hotels, concert and sporting events, places of worship, and even public restrooms. China uses it to dispense toilet paper in public restrooms to monitor usage and avoid toilet paper theft. I guess it’s all part of the experience.
The technology is also being evaluated for its effectiveness with emotional awareness. Some of the well-known tech companies are developing software that can recognize and detect external emotions based on facial expressions. The technology can ascertain facial muscles to determine whether you are happy, sad, angry, confused, tired, calm or fearful, and may even be considered for predictive behavioural outcomes (i.e. falling asleep at the wheel, or committing crimes). Facial recognition technology can only of course detect the external emotions and not what a person may internally feel so it must be used with caution.
Recognising the Challenges
As with many emerging technologies, a lack of standards and/or regulatory controls can make users feel insecure. This applies to facial recognition technology that currently operates without federal regulations. While it is improving, the technology is certainly not 100% accurate and can lead to bias and misleading results. As a surveillance tool, the technology has been known to misidentify people based on their social identify (i.e. gender and race). This can lead to significant abuse of use whether it be tied to civil right or other privacy issues. Accuracy and accountability are important in the use of facial recognition, starting off with the legal and judicial systems.
When it comes to what may be considered invasive surveillance along with potential data hackers, there is no question that facial recognition creates warranted concern for people’s security and privacy. With the heavy focus on privacy issues, similar to other AI technologies, facial recognition will require careful regulatory controls on usage to avoid further abuse. Facial recognition software can be a powerful and effective technology when implemented correctly but can also cause irreparable harm with privacy and security.
Future for Facial Signatures
The human face is a prime location for identification, and with advancements in AI and real-time data transmission is opening up numerous opportunities across law enforcement, health and financial services, transportation, retail, marketing, social media and other services. The confidence in use across any of these industries is contingent on security.
Yes, my face is now my signature password. There is no turning back. I just hope I can make my plane on time.
Are you a Distributor, Vendor, Retailer, Reseller or Channel Services Provider? Don’t miss your chance to win a PCR Award! Submit your entries for free here before 6th December!
Read the latest edition of PCR’s monthly magazine below: