Let us start with a simple question; what is facial recognition? Simply put, facial recognition recognizes faces by analyzing facial features. In this article, we will discuss how, why and when to use facial recognition. To be able to successfully use facial recognition the software has to have some kind of database to compare facial features. To make this easier to understand we’ll use an example. You are a criminal, you have an outstanding warrant and you are driving across Brooklyn Bridge to meet someone in Manhattan. When you enter the bridge, you drive under cameras that take a picture of every single car and driver. You, as a US citizen, have a picture of you in a national database. After a few minutes you are across the bridge and a posse of police cars are gently awaiting your arrival. How you may ask? When you drove under the cameras, they recognized your facial features, saw that you had an outstanding warrant and contacted the police. This is just one of the uses of facial recognition that will be in wide use in the next few years. Why? It is efficient and saves a lot of money in the long run.
Let us move on to the pros and cons.
Facial recognition can work wonders when it comes to preventing crimes and if a crime should happen it will be significantly easier to pursue the criminal due to the fact that the private security company or government official will have a full overview over which individuals that have been in the targeted area.
Discovering sickness and hygiene
Facial recognition is not just for the public. It can also act as a security measure for companies and alike. Being analyzed every day can track your skin tone and such alike. This can help discover sickness early on and potentially save lives. Using facial recognition instead of for example ID cards and fingerprint scanners is much safer and more hygienic. There is no touching and there is a smaller chance that you will lose your face compared to a small ID card.
Video surveillance requires A LOT of cameras. For a large country, we are talking tens of millions of cameras, if not even more. Someone has to produce all of these and set them up. If you count in all the extra equipment that is necessary in many cases, like infrared illuminators to light up dark spaces, etc. In short, there is a lot of work to get national surveillance up and running.
To use large scale facial recognition successfully you will need unspeakable amounts of data. This has several side effects. For one there is an increasing demand for data centers with a lot of equipment. This creates a lot of jobs. The increasing amount of data also makes for a bigger carrot for the compression companies (HBO’s Silicon Valley for example).
Recognizing people with certain kinds of skin color
One of the most logical reasons for this is that the public photographs that the tech-companies use for training their AI have more people with light skin than dark skin. This causes the software to get better at recognizing light-skinned people than dark-skinned. The other reason we shall bring up is the lack of contrast in dark skin and how some women cover up their facial features with varying degrees of makeup. This confuses the AI and gives it too little information to work with for successful recognition.