Facial Recognition.

Unlock with a look.

The smartphone industry has seen pretty much major change in 2018.

• The facial recognition, Bezel-less displays, In-Glass fingerprint scanner and a lot more technology stuffed into the mobile phones we carry today.

The human face has an astonishing varieties of features which not only help us recognize others but read and understand them through a constant for flow of intentional and unintentional signals. It’s one of the unique functions that separates man from machine until now.

Photo by Aatik Tasneem on Unsplash

Fact: Technology developed by Facebook’s AI can now recognize faces with 97.37% accuracy which is 0.28 % less accurate than a human.
(Human — 1, Artificial Intelligence — 0) which is surprising because computers are more accurate than us.

How do computers recognize our faces??

Initially, the computer would divide the face into landmarks or nodal points which are pretty much the depth of the eye sockets, the distance between the eyes, the width of your nose and length of your lip.

Everyone has different coloured eyes, nose structure, lips and unique ears.
So the measurement of the nodes is different for everyone and hence the password.

• All these information are made into a unique code and we call it the person’s own “ faceprint”. But there is a problem, to get a correct match each and every time your photos has to be the same or like to like.
• Our faces are in constant flux and they are not just static like fingerprints.

How does this technology work?

Every face has numerous, distinguishable landmarks, the different peaks and valleys that make up facial features. Each human face has approximately 80 to 100 nodal points.

• It scans faces and measures distinguishing facial features such as eye position, eyebrow shape, and nostril angle. This creates a distinctive digital “faceprint” — much like a fingerprint — which the system then runs through a database to check for a match.

Facial Recognition Technology scans:

1. The distance between the eyes
2. The width of the nose
3. The depth of the eye sockets
4. The shape of the cheekbones
5. The length of the jawline

These nodal points are measured creating a numerical code, called a Faceprint, representing the face in the database.

Nodal points

The Four stage process that the system does.

1. Capture — a physical or behavioral sample image is captured by the system during enrollment ( Find a face in the image).
2. Extraction — unique data is extracted from the sample and a template is created. (Analyse facial features)
3. Comparison — the template is then compared with a new image (Compare the image with the database.)
4. Matching — the system then decides if the features extracted from the new sample are matching or not (Make a prediction)

But we have some problems.

We face 4 main issues with facial recognition.

The A-PIE Problem

Ageing, Pose, Illumination, Emotions.

• To solve this problem we have 3D recognition system called, DEEPFACE.
  It’s able to take a 2D photo of a person and create a 3D model of the face.

Apple’s Dot projector doing its 3D model work. — Apple

So now it scans from any angle or poses can be compared.
So this solves Pose from the A-PIE problem.

• Ageing is no longer a problem either. The faceprint system is now redefined to capture the areas of the face that have rigid tissue like the curvature of the jaw, the forehead doesn’t alter too much in the course of time.
• To deal with Illumination, mobile phones are equipped with IR blaster near the front facing camera (Mi 8's IR sensor, Apple’s Flood Illuminator).
• Emotions are dealt by the system by learning human emotions. This is done by Deep Learning.

Emotion — Photo by Ryan Franco on Unsplash

So the more you use the facial recognition on your phone the better it gets over time.

In upcoming years there’s gonna be a Surface Texture Analysis, where the device recognises the texture of the skin rather the facial details.
This technology can identify between identical twins with their skin texture.

No doubt the fruit company is gonna bring it to the world. (“wink wink”).

PS : There will future updates on this post if something new turns up.

Follow for updates

SS

Thanks for reading.

Until next time

Peace, Love and Gratitude.