+1 386-243-9402 MON – FRI : 09:00 AM – 05:00 PM

Faces by John R. Patrick

Written: March 2023


Now that I have covered cats and dogs with machine learning, I will branch out to facial recognition of humans. Facial recognition is a technology capable of identifying or verifying a person from a digital image or a video frame from a video camera. Facial recognition has also become an alternative for fingerprints as a form of biometrics to make smartphones more secure.


In September 2017, Apple announced Face ID. This provides some insight about how facial recognition technologies work. Apple said, “With a simple glance, Face ID securely unlocks your iPhone X.”   With Face ID, you can make purchases from Apple and make payments with Apple Pay. Apps which support Touch ID will automatically support Face ID, and many more will surely follow. The technology behind Face ID is quite impressive. The camera captures and analyzes more than 30,000 invisible dots projected onto your face plus an infrared image of your face. The data is compared to your faceprint, the data gathered from you when you looked at your phone from different angles when you set up the Face ID feature. Face ID works in the dark and can adapt to shaving a beard, and wearing a hat, scarf, or sunglasses. While the odds of someone stealing your iPhone and having a fingerprint just like yours is 1 in 50,000, with Face ID, the odds of the thief having a face the same as yours is 1 in a million.  From a privacy perspective, Apple’s facial recognition is between you and your iPhone. Nothing is stored in the cloud. Google’s version of Face ID is called Face Unlock. (Google’s Bard AI said it is not as secure as Apple’s Face ID. Now let’s take a closer look at how facial recognition works.


Facial recognition technology has been around for decades, but it has become a more practical application as a result of dramatic increases in the processing power of computer servers and mobile devices and the vast amount of low-cost storage available. Facial analysis capabilities, such as those available in Amazon Rekognition, allow users to be aware of faces which exist in an image or video, and what attributes those faces have. For example, software could analyze attributes such as eyes open or closed, mood, hair color, and the visual geometry of a face. Although around for a long time, awareness and accessibility of the technology have increased greatly. During the past few years, it has become an integral part of powerful and innovative solutions, such as personal photo applications and secondary authentication for mobile devices. To understand these emerging capabilities, let’s first discuss more about how facial recognition works.


Every face has numerous, distinguishable identifiers. For example, FaceIt, a London based company founded in 2012 and a provider of a technology platform for multiplayer video games, defines these identifiers as nodal points. Each human face has approximately 80 nodal points. Examples of the nodal points measured by FaceIt software are,

  • Distance between the eyes
  • Width of the nose
  • Depth of the eye sockets
  • Shape of the cheekbones
  • Length of the jaw line

These nodal points and others are measured and then converted into a numerical code called a faceprint, which represents the face in the database. The faceprints, in combination with other attributes can enable an organization to categorize and organize or search through millions of images in seconds. For example, an organization can create a query to look through a large database of images to find an elderly person with glasses and brown hair who has a certain sized nose and shape of cheekbone. The query could also be submitted with an image taken by a surveillance camera and then look for a person in the database who matches the query.


These detected attributes become increasingly useful for customers needing to organize or search through millions of images in seconds using metadata tags (e.g., happy, glasses, age range) or to identify a person (i.e., facial recognition using either a source image or a unique identifier). Marinus Analytics is a woman-owned company, founded in 2014 as a spin out from Carnegie Mellon Robotic, which builds AI tools to turn big data into actionable intelligence. Using AI such as the facial recognition features available in Amazon Rekognition, the company is dedicated to helping find human trafficking victims and reuniting them with their families. Emily Kennedy, President and Co-Founder of Marinus Analytics, said,


We believe that artificial intelligence shouldn’t be feared by humans because it can enable us to accomplish far more good than would otherwise be possible and scale the overall impact. For Marinus Analytics, we have been lucky enough to see lives saved by using AI technology because it has empowered those fighting against human trafficking to accomplish seemingly impossible tasks in the face of what initially seemed like an insurmountable pile of data.


Aella Credit is a financial services company based in West Africa which provides instant loans to individuals with a verifiable source of income in emerging markets by using facial recognition. One of the major challenges to providing banking services to those in emerging markets is identity verification and validation for people who don’t have easy access to retail banking services. Aella Credit uses Amazon Rekognition for biometric identity verification on a mobile application. Using Rekognition in the application allows customers to verify their identity and get access to banking services with minimal friction. The process is exactly like I will describe in a follow-on article for Coinbase. Aella Credit enables customers to upload a photo of their government-issued ID, and then take a selfie for verification. Aella first verifies the government-issued ID against the government database, and then uses Amazon Rekognition to compare the two images to see if they are a match.


Wale Akanbi, CTO and Co-Founder of Aella Credit, said,


The ability to properly identify users is a key hindrance in building credit for billions of people in emerging markets. Using Amazon Rekognition for identity verification on our mobile application has reduced verification errors significantly and given us the ability to scale. We can now detect and verify an individual’s identity in real time without any human intervention, thereby allowing faster access to our products. Amazon Rekognition helped us effectively recognize faces of our customers in our markets. It also helped us with KYC [know your customer] in discovering overlapping profiles and duplicate datasets.


Facial recognition with AI offers many benefits, but there are also many risks to be considered. Emily Kennedy, quoted above saying, “artificial intelligence shouldn’t be feared by humans” may now feel differently. Today, along with 1,000 others, I signed a letter written by Elon Musk, CEO of SpaceX, Tesla & Twitter and Steve Wozniak, Co-founder, Apple. The letter, published by futureoflife.org, is a plea to ‘Pause Giant AI Experiments‘. I’ll have more to say about this in another article. Next up some more about face recognition.