How easy is it to identify a person you have met before? Usually, very easy! You may say it is because you remember him. Technically, your brain is matching it against a memory it has. Face recognition by machines work in a similar fascinating way.
The Dictionary definition of face recognition is the ability of a computer to scan, store, and recognize human faces for use in identifying people.
Today this term is not very alien. Apple users use Face ID to unlock their phones. Airport security uses it to identify convicts. Workplace safety apps like Truein use it to mark attendance or authorize visitors.
There are many variations of this technology. We will look at the most common approach. Let’s break down the definition.
1. Scan: A person walks by a machine that takes its picture and identifies its geometry. By geometry we mean distance of the nose from chin or location of eyes. There can be 60 to 80 such data points that make the complete face ID.
2. Store: The face ID, a mathematical formula consisting of data points is stored in the database. The computer understands this as your signature – unique to you.
3. Recognize: Next time you walk past the same machine, it can create a similar map and trace it against the database it has. Result? You are identified.
Face recognition was first invented by Woodrow Wilson Bledsoe in 1960s. Back then the degree of errors was vast and the amount of time it took to return a result was huge. Part of it was due to computing power of machines back then.
Besides, the number of facial markers used to create a Face ID were low. In 1970s there were 21 facial markers. Today, the sophisticated machines use as many as 80, increasing accuracy.
Facial recognition gets seriously compromised if the machine is not geared to identify in low-light or with changing facial features. It will result in two things:
It is therefore critical that you pick a face recognition technology that stands the test of time. A time-tested and proven technology by a trusted provider is the key to adopting this futuristic technology.
We believe that accuracy has to be practically hundred percent. Any small error leads to manual work, beating the purpose of our technology. This hundred percent perfection has happened over many years through many iterations. Today, we are proud to meet those benchmarks.
Biometric IDs are unique and beat the traditional paper-pen or card systems hands down. Cards can be swiped by proxies and attendance can be fudged. It is a common malpractice, often seen at plants within contract workers.
The thumb IDs are biometric too, but they have failed to meet the hundred percent benchmark. Women using Indian Mehndi on hands feel frustrated when their thumbs don’t get identified at office machines. Similarly, plant workers who work on oily machines cannot walk past a thumb-based system easily.
Besides, it is not touchless, and it takes multiple attempts for one correct match. Face recognition is truly the only unique identifier, doing its job in few seconds. It cannot be proxies or scope for manual overrides.
Contract and Distributed workforce Time & Attendance has these age-old problems! !
Find out how to solve them.
