There has been a lot of press coverage about facial recognition software over the last few years, and if there is one sure thing, it is here to stay. We are seeing its use increasing, and not only in the form of CCTV cameras watching us in public places or places of business. You will find that there are many applications for this technology which you can see on your Smartphone and in companies throughout the world. Below is a summary of how this technology works so you may consider incorporating it into your business.
How Facial Recognition Works?
For the software to recognise a face, the details will first need to be submitted, and you can do this through a photo or video of a person. You may have a picture in the person’s profile view, or a straight view of the face looking directly at the camera. When the image is complete and uploaded, it is time for the software to do its work and map the face’s features to allow the face recognition software to work correctly.
Mapping The Face
The software’s algorithm will take the face and then map the facial features and geometry of the face, which it will use as a reference. There are key factors that the software will map, such as:
- The distance between the eyes.
- The distance between the forehead and chin.
- The length and width of the nose.
- The shape of the eyes.
- The cheekbone structure.
The software will map what are called nodal points, and it will create around 80 or more of these nodes for each face. With the information added to the system, it can then identify a specific person with a high degree of accuracy.
A Specific Mathematical Formula For Each Face
When the data upload is complete, the software will create a unique mathematical formula for each face that is stored. The formula can then be called on when searching for the person you are looking for, and the closest matches will show in the results.
Deciding On A Match
When the software has finished running, it will give you a score with a percentage of probability for a match. The accuracy of facial recognition is improving all the time, and one average it increases in efficiency by about 50% every two years. It is vital to keep in mind that it is not 100% accurate, and there are ways that people can try and cheat the software. Still, as a whole, it is suitable for many different applications, such as a security feature on a mobile phone, at airports and government buildings, and social media.
