What is Facial Recognition?
Facial or body recognition involves identifying or verifying someone using characteristics and features of their face or body shape. Facial recognition has been in use since 1964 and has been improving over the years. Facial recognition was first pioneered by Woody Bledsoe, along with Charles Bisson and Helen Chan Wolf and their work was funded by an anonymous intelligence agency. A lot of their work was never made public.
On average, the error rate has reduced by half every two years. Over the decades, this technology has taken several leaps and bounds resulting to better performance and practical uses. Apple uses facial recognition to unlock your Iphone and authorise payments in Apple Pay just to name a few use cases. Facebook also uses facial recognition. You receive notifications when a photo of you is uploaded by one of your friends. You can either approve or hide these photos in your news feed. Facebook provides information on how you can opt-in or out of face recognition here.
How does Facial Recognition work?
Facial and body recognition uses algorithms to identify certain points on the face and body, such as the shape of one’s chin, arms and legs outline and create a data template for that person. It captures, analyzes, and compares patterns based on the person's facial and body details. To capture this data, Paparazzo has a convolutional neural network. CNN, or ConvNet is a class of deep neural networks, most commonly applied to analyzing visual imagery. Paparazzo transforms the analog information (a face) into a set of digital information (data) based on the person's facial features. CNNs take advantage of the hierarchical pattern in data and assemble patterns of increasing complexity using smaller and simpler patterns embossed in filters. If this data is in the shape of a human it will then lock onto the subject and start to follow it. Interestingly we have also found it can capture teady bears.
Convolutional networks were inspired by biological processes in that the connectivity pattern between neurons resembles the organization of the animal visual cortex. Individual cortical neurons respond to stimuli only in a restricted region of the visual field known as the receptive field. The receptive fields of different neurons partially overlap such that they cover the entire visual field. CNNs use relatively little pre-processing compared to other image classification algorithms. This means that the network learns to optimize the filters or convolution kernels that in traditional algorithms are hand-engineered. This independence from prior knowledge and human intervention in feature extraction is a major advantage.
People have said that facial recognition is “the most natural of all biometric measurements” and say the technology is highly accurate, with “vanishingly small differences" in their rates of false-positive or false-negative readings. Considering we recognise people with our face, and Paparazzo uses technology that resembles the way a human eye works, this can only be a good thing.
What is Artificial Intelligence?
Artificial Intelligence or AI as commonly referred to involves a broad range of computer science which concerns itself with building smart machines or devices to perform a specific task. Your latest mobile phones, be it Samsung S20, Iphone 12 Pro Max, and even its younger versions use AI in their native camera app. It can make decisions on the brightness level, identify the subject in the foreground versus the background to achieve the background blur for your profile photos and videos, and even make complex calculations so your night photos still come out sharp with good highlights and lowlights.
How Facial Recognition and AI is used.
When these two technologies come together, the products or service it can come out of it is nothing short of powerful. We have seen both technologies used for Filming, Video Calls, Vlogs and even Sports.
We have listed some of our favorite use cases below, we also see some amazing examples in sport.
- Airports enable facial recognition for passengers boarding flights. Travellers' faces can be scanned by a camera to have their identity verified to board their plane without showing their passport or boarding pass.
- Healthcare providers allow customers to file health insurance claims which are signed using a photo, rather than a written signature, in a bid to cut down on instances of fraud. Other examples are in hospitals where customers can check in using their face and also apps that make sure customers are taking their prescriptions.
- Brick and Mortar Stores use facial recognition technology allowing customers to virtually "try on" make-up using in-store augmented reality mirrors or offering ads to people based on their facial expression. Soon we might be able to try on clothes virtually.
- Social Media filters were one of the pioneers of facial recognition software: it allows brands and organizations to create filters which mold to the user’s face — hence the crazy dog faces and aliens we see today. TikTok has taken this to a new level.
- Hawk-eye, as used in Tennis grand slam tournaments, is a great example how both these technologies come together. Hawk-Eye uses six or more computer-linked television cameras situated around the court. The computer reads in the video in real time, and tracks the path of the tennis ball on each camera. These six separate views are then combined together to produce an accurate 3D representation of the path of the ball.
- VAR or Video Assistant Referee in the Football world rely on video footage taken using high speed smart tracking cameras positioned around the stadium and in several angles.
The future of Facial Recognition
Within the consumer technology space, this technology is more ubiquitous with Smart Phone Holder Trackers. It works similar to Hawk-Eye, but instead of tracking the tennis ball, it will track the subject using facial recognition and make accurate decisions which direction to track because of AI. This is in largely in part due to the advances in deep convolutional neural networks – which is used primarily to classify images and perform object recognition.
Facial Recognition FAQ
How accurate is Facial Recognition?
Face recognition can be 99.97% accurate. However, depending on the conditions, poor lighting and positioning, algorithms can have a 9% mismatch rate or higher. Paparazzo can sometimes find it hard to track individuals if they are camouflaged into the background. For example, wearing green in the middle of a forest and it would struggle to find the outline of the human being.
Is AI Facial Recognition Safe?
Unlike other identification solutions such as passwords, verification by email, selfies or images, or fingerprint identification, biometric facial recognition uses unique mathematical and dynamic patterns that make this system one of the safest and most effective ones. Paparazzo does use any app or Bluetooth connection in its technology. It is all built into the little camera inside the unit. Therefore not only is the technology safe, you know it's not using your data for anything else or sending it to some undercover government somewhere.
How does Facial Recognition work if you are wearing a mask?
Throughout 2020, face masks were “breaking” facial recognition systems – after all, they blocked out a large portion of the subject’s face. We have successfully used Paparazzo to identify and track people when they have been wearing a mask as it does not solely rely on the face to track the subject.
