The concept of facial recognition became a buzzword when Apple introduced its Face ID unlock feature in 2017. This fascinating implementation of image analysis has it’s applications ranging from marketing to security. On top of that, the industry once valued at $4 billion in 2017 is now expected to grow to $7.7 billion by 2022. So how does this piece of engineering work and is it reliable?

            So how are these recognition systems built? Well object detection is primarily based on computer vision and image processing techniques. The primary objective of any face detection algo is to state if there is a face present in the picture. Face detection uses methods feature based, appearance based, knowledge based and template matching. Furthermore, appearance-based method uses eigenface based, distribution based, neural networks and much more complicated algorithms to process the image. The feature based method detects faces by trying to map out our structural facial features and has a success rate of 94%.

Now that similar technologies are being implemented by law enforcement and for personal safety, there have been quite some privacy concerns. As similar technologies keep getting popular, your facial signature might end up someplace else without your knowledge. Since these systems aren’t a hundred percent accurate, there might be cases of mistaken identity since law enforcement is also using the technology. Similar questions continue to prevail as recognition systems keep improving.

So far, facial recognition systems have been the most promising implementations of computer vision and image processing. This new technology has only been in use for a few years and yet it has been able to take over the market like a wildfire. Our dependence on ML keeps growing daily and the pandemic added to it. As these algorithms get more accurate, the future of recognition systems looks bright.