Facial recognition – Science fact or science fiction?

Facial recognition is becoming an ever-present part of our lives in a variety of different ways. Technology has developed to a point where what was once only seen in the movies is now very much a reality. From instantaneously unlocking our mobile phones to facial recognition processing at passport control – we can now be identified via our faces with incredible precision and accuracy.

We thought we’d look into how this came to be, how the technology has developed, what is currently possible and what the future holds for facial recognition.

When was facial recognition invented?

Facial recognition was first developed in the 1960s by Woody Bledsoe, Helen Chan Wolf, and Charles Bisson for a US military agency. All three had slightly different backgrounds in various areas of mathematics, robotics and computer science but those complementary skills enabled them to develop a method where a computer program could identify faces by measuring the distances between landmarks on a face.

The next significant breakthrough came in 1997 from a team at the University of Bochum who developed software called ZN-Face which was adopted by companies around the world for use in commercial applications. Since then the technology and algorithms have evolved at an incredible rate but full credit belongs to those original pioneers of Bledsoe, Wolf and Bisson.

How does facial recognition work?

As previously mentioned, when facial recognition was first developed it worked by calculating the distances between landmarks on a face (though these had to be manually identified in the first instance) and matching them against records in a database. Nowadays there are a variety of different techniques and solutions being used, some of which include:

Geometric and photometric

Traditional facial recognition programs still work by identifying landmarks on an individual’s face with a variety of different measures, be it position, size, distance etc. They then usually have either a geometric approach or a photometric approach. The geometric approach, in a sense, turns the face into a geometric representation calculating distances between the features while a photometric approach works with the pixels of an image assigning them values which can then be matched.

3D recognition

3D recognition is a highly sophisticated method which uses sensors to scan the 3D elements of a face giving you an extremely unique set of data. It’s extremely effective when it comes to different viewing angles given it has a 3D model from which to compare and analyse. If you’re in a situation which allows you to have multiple cameras set up in a single space (like many do in security settings) then you can use this method to track and identify individuals in real-time.

Skin texture recognition

The fast-paced development of technology has allowed for some very specific and nuanced development approaches such as skin texture recognition, which looks at patches of skin on an individual’s face, identifying unique features such as wrinkles, pores, spots etc. and creates a unique identifier from these. This approach is still in development and has so far been used in conjunction with more traditional facial recognition to improve the overall accuracy of results.

Thermal recognition

Another completely different method that has been tried and tested (and is still being developed) is through thermal recognition. Using thermal cameras, one of the main advantages of this approach is its effectiveness in low-light and/or night-time scenarios where normal cameras might struggle to capture a workable image. As with skin texture recognition, this usually works in conjunction with conventional facial recognition techniques to improve the precision of the overall results. 

What is facial recognition used for?

The most common uses of facial recognition include a variety of different purposes from entertainment to advertising and security. Smartphones are the most obvious and widespread piece of technology that has incorporated facial recognition as an unlocking feature, which has quickly become a standard feature.

Mobile apps also take advantage of this capability in all sorts of creative ways through social media apps that can manipulate your photos/create avatars, games, banking apps etc. With respect to advertisers, screens with facial recognition are being used in retail spaces around the world to run targeted ads based on the faces of people that are in the area.

Last but not least, facial recognition has become sophisticated and precise enough to now be used in security situations commercially, domestically and internationally, cross-referencing databases with thousands, hundreds of thousands and sometimes even millions of subjects.

All these uses were in the realm of science fiction in the 1960s when Bledsoe and co were first developing the technology, but now they are very much science fact.

What will facial recognition look like in the future?

With facial recognition firmly established, the next question is what will it look like in the future? Given we now have working examples, we can make predictions about what the future holds.

One thing certain is that facial recognition will become more prevalent in our lives with growth in the industry only heading in an upwards direction. A recent report by Market and Markets predicts that global facial recognition will be valued at $7 Billion USD by 2024, twice what it is now.

Obvious advances will come as a result of improved speed, accuracy and scale of facial recognition, all of which will lead to their own and cumulative developments. One example is the further development of branches within facial recognition such as expression recognition.

This is an area that focuses on analysing the non-verbal communication cues expressed by a subject at any point in time. This will be highly applicable in contexts such as advertising, for example, where businesses will be able to detect how favourably users are reacting to their advertisements or promotions.

Another outcome we can anticipate is the increase in the mobility and reach of facial recognition. Government, military and security organisations continue to invest in facial recognition increasing the capacity for real-time monitoring of larger crowds and higher transit areas.

Drones have already been deployed to assist fixed cameras in this way as well as functioning on their own which means facial technology will be able to be deployed to match the range of a drone. While the military implications here are fairly obvious, law enforcement and police are already making plans to use such drones to find missing and vulnerable persons.

The uptake and accessibility of facial recognition as the technology becomes more cost-effective will also mean it becomes more heavily integrated into our lives and could well become the dominant biometric identifier upon which society depends.

Where Science Fiction becomes Science Fact

As you can see, Facial Recognition is a fascinating area that will see tremendous growth and development now and in years to come. The most exciting part is that as much as one attempts to predict and anticipate what will happen, there will, without doubt, be innovations that will take everyone by surprise, leading to new possibilities.