Facial recognition: top 7 trends (tech, vendors, use cases)
Face recognition – fascinating and intriguing
Few biometric technologies are sparking our imagination quite like facial recognition.
Equally, its arrival has prompted profound concerns and surprising reactions in 2020.
But more about that later.
In this web dossier, you will discover the seven face recognition facts and trends set to shape the landscape in 2021.
Top technologies and providers
AI impact - Getting better all the time.
2019-2024 markets and dominant use-cases
Face recognition in China, India, United States, EU, and the UK, Brazil, Russia...
Privacy vs. security: laissez-faire or freeze, regulate, or ban?
Latest hacks: can facial recognition be fooled?
Towards hybridized solutions.
Let’s jump right in.
How facial recognition works?
Facial recognition is the process of identifying or verifying the identity of a person using their face. It captures, analyzes, and compares patterns based on the person's facial details.
The face detection process is an essential step in detecting and locating human faces in images and videos.
The face capture process transforms analog information (a face) into a set of digital information (data or vectors) based on the person's facial features.
The face match process verifies if two faces belong to the same person.
Let’s illustrate this 3-step process with a recent example.
A student from the greater Washington DC area used an open-source facial extraction app to detect and deduplicate over 6,000 images of faces from 827 videos posted on Parler during the 6 January event outside and inside the Capitol building He created a website called Faces of the Riot, where these portraits are displayed.
Demonstrators, rioters, and journalists have done part of the face capture step with their smartphones (analog face to the digital picture).
He used facial detection to extract faces from 200K images.
It’s up to the FBI to investigate, transform the portraits (digital pixels to vectors) and potentially do the face match with existing databases and identify the individuals (with an AFIS / ABIS system).
Today it's considered to be the most natural of all biometric measurements.
And for a good reason – we recognize ourselves not by looking at our fingerprints or irises, for example, but by looking at our faces.
View Complete Report On Facial System By DLR
Before we go any further, let's quickly define two keywords: "identification" and "authentication."
Face recognition data to identify and verify
Biometrics are used to identify and authenticate a person using a set of recognizable and verifiable data unique and specific to that person.
For more on biometrics definition, visit our web dossier on biometrics.
Identification answers the question: "Who are you?"
Authentication answers the question: "Are you really who you say you are?"
Stay with us. Here are some examples :
In the case of facial biometrics, a 2D or 3D sensor "captures" a face. It then transforms it into digital data by applying an algorithm before comparing the image captured to those held in a database.
These automated systems can be used to identify or check an individual's identity in just a few seconds based on their facial features (geometry): spacing of the eyes, bridge of the nose, the contour of the lips, ears, chin, etc.
They can even do this in the middle of a crowd and within dynamic and unstable environments.
Owners of the iPhone X have already been introduced to facial recognition technology.
Of course, other signatures via the human body also exist, such as fingerprints, iris scans, voice recognition, digitization of veins in the palm, and behavioral measurements.
Why face recognition, then?
Facial biometrics continues to be the preferred biometric benchmark.
That's because it's easy to deploy and implement. There is no physical interaction with the end-user.
Moreover, face detection and face match processes for verification/identification are speedy.
So, what is the best face recognition software?
#1 Top facial recognition technologies
In the race for biometric innovation, several projects are vying for the top spot.
Google, Apple, Facebook, Amazon, and Microsoft (GAFAM) are also very much in the mix.
All the software web giants now regularly publish their theoretical discoveries in artificial intelligence, image recognition, and face analysis to further our understanding as rapidly as possible.
Let’s take a closer look :
Academia
The GaussianFace algorithm developed in 2014 by researchers at The Chinese University of Hong Kong achieved facial identification scores of 98.52% compared with the 97.53% achieved by humans. An excellent rating, despite weaknesses regarding memory capacity required and calculation times.
Facebook and Google
In 2014, Facebook announced its DeepFace program, which can determine whether two photographed faces belong to the same person, with an accuracy rate of 97.25%. When taking the same test, humans answer correctly in 97.53% of cases, or just 0.28% better than the Facebook program.
In June 2015, Google went one better with FaceNet. On the widely used Labeled Faces in the Wild (LFW) dataset, FaceNet achieved a new record accuracy of 99.63% (0.9963 ± 0.0009).
Using an artificial neural network and a new algorithm, the company from Mountain View has managed to link a face to its owner with almost perfect results.
This technology is incorporated into Google Photos and used to sort pictures and automatically tag them based on the people recognized.
Proving its importance in the biometrics landscape, it was quickly followed by the online release of an unofficial open-source version known as OpenFace.
Microsoft, IBM, and Megvii
A study done by MIT researchers in February 2018 found that Microsoft, IBM, and China-based Megvii (FACE++) tools had high error rates when identifying darker-skin women compared to lighter-skin men.
At the end of June 2018, Microsoft announced that it had substantially improved its biased facial recognition technology in a blog post.
Amazon
In May 2018, Ars Technica reported that Amazon is already actively promoting its cloud-based face recognition service named Rekognition to law enforcement agencies.
The solution could recognize as many as 100 people in a single image and can perform face matches against databases containing tens of millions of faces.
In July 2018, Newsweek reported that Amazon’s facial recognition technology falsely identified 28 US Congress members as people arrested for crimes.
Key biometric matching technology providers
At the end of May 2018, the US Homeland Security Science and Technology Directorate published the results of sponsored tests at the Maryland Test Facility (MdTF). These real-life tests measured the performance of 12 face recognition systems in a corridor measuring 2 m by 2.5 m.
Thales' solution utilizing a Facial recognition software (LFIS) achieved excellent results with a face acquisition rate of 99.44% in less than 5 seconds (against an average of 68%), a Vendor True Identification Rate of 98% in less than 5 seconds compared with an average 66%. It also achieved an error rate of 1% compared with an average of 32%.
March 2018 – The live testing done using more than 300 volunteers identified the best-performing facial recognition technologies.
More on performance benchmarks: The NIST (National Institute of Standards and Technology) report, published in November 2018, details recognition accuracy for 127 algorithms and associates performance with participant names.
The NIST Ongoing Face Recognition Vendor Test (FRVT) 3 performed at the end of 2019 provides additional results. See NIST report.
NIST also demonstrated that the best facial recognition algorithms have no racial or sex bias, as reported in January 2020 by ITIF. Critics were wrong.
In NIST's reports (August 2020 and March 2021) entitled "Face recognition accuracy with face masks using post-COVID-19 algorithms", we see how algorithms, in less than a year, are increasing their performance.
Facial Emotion Recognition (FER)
Facial Emotion Recognition (from real-time or static images)is the process of mapping facial expressions to identify emotions such as disgust, joy, anger, surprise, fear, or sadness - or compound emotion such as sadly angry - on a human face with image processing software.
There are also three steps in the recognition or interpretation of human emotions:
1) Face detection
2) Face expression detection
3) Assignment of expression to a specific emotional state.
Facial emotion detection's popularity comes from the vast areas of potential applications.
It's different from facial recognition, whose goal is to identify a person, not an emotion.
Face expression may be represented by geometric or appearance features, parameters extracted from transformed images such as eigenfaces, dynamic models, and 3D and models.
Providers include Kairos (face and emotion recognition for brand marketing), Noldus, Affectiva, or Sightcorp.
