What is Facial Recognition?
When we see someone we know our brain process a huge of amount of data about them, the data include from the pattern of hairs, the colour of skin to the shape of eyes, lips, teeth, ears, neck etc. which our brain matches it with its database to tell you who’s that person is.
A facial recognition is in its simplest form do the same thing as our brain to recognise the face. It is a technology that is built to recognize your face. when you point your phone’s camera to face of one or more people, it will assume it’s a face and stick a box around it, track it and make sure that it will remain in focus, if you point your camera to a crowd it will identify multiple faces The facial recognition or the Biometric Artificial Intelligence are trained to uniquely identify the faces by analysing pattern, shape & textures of your look like your eyes, nose, teeth, smile etc.
The advanced facial recognition systems not only billboard your face but also capable of finding out who a particular face belongs to.
How does it identify you?
Each face has around 80 nodal points or landmarks. The Present Facial Recognition works on Artificial Intelligence algorithms which are designed to read that landmarks of your faces even if you are in a crowd of millions. Some of the nodal points of the face are Distance between the eyes, Width of the nose, depth of the eye sockets, The shape of the cheekbones, The length of the jawline etc.
These nodal points are measured creating a facial signature which is a numerical code that represents the face in the database. A probe image when required is compared with the stored facial signature data.
Techniques used in Face Acquisition:
The face Acquisition process performed in two steps, the first one involves the extraction & selection and second involve classification. Recognition algorithms are broadly divided into four main approaches.
- 2 Dimensional Recognition
- 3 Dimensional Recognition
- Skin Texture Analysis
- Thermal Sensing Technology
2D or Geometric Approach
The geometric approaches look into separate structures of the face and these geometrical structures of the face are extracted from eyes, the shape of the mouth, face boundary etc. and are organized as a graph for modelling and recognition.
However, these algorithms can be classified into two broad categories: holistic based and feature-based models.
The Holistic based model recognises the face in its entirety while the feature-based model subdivides the subject face into components such as according to features and analyse each as well as its spatial location concerning other features.
The various face detection techniques used in this approach are Principal Component Analysis (PCA), Neural Networks, Machine Learning, Hough Transform and Template Matching.
Recognition using 3D Approach
In this type of approach, 3Dimensional geometry of human face is used to achieve better results than 2D. 3D facial recognition techniques use distinguished topographies of the face which are not changed with time. They are stiff tissue and bones, such as the curves of the eye socket, nose and chin — to identify the subject. These are all unique to every human and don’t change with coarse of time. Further-more to know that this technique is also not affected by a change in the quantity of light and the face can be identified from different view angles.
These 3D recognition techniques used 3D image sensors to capture face geometry. The Sensors are placed on CMOS Chip which is designed to hold dozens of sensors. The sensors throw a sophisticated structured light on the face which was again collected by the sensors. Each sensor captures a different part of the spectrum which was collected and processed to generate the 3D image of the subject.
Skin Texture Analysis
Our skin has the same kind of details like our fingerprints. The technique called skin textures analysis uses the visual details like unique lines, patterns and spot of the skin and convert into a digital data.
This process is called surface texture analysis in which a picture is taken of a particular area (patch) of a skin the picture is then converted into digital blocks, algorithm further used to identify each pattern by distinguishing any lines, pores and the actual skin texture. It can also able to detect the difference between the two identical twins which was not possible with facial recognition software alone.
Tests have shown that with the addition of skin texture analysis, performance in recognizing faces can increase 20 to 25 per cent.
Thermal Sensing Technology
Another technique is thermal Sensing Technology which uses the subject temperature to produce an image. Thermal imagers work by distinguishing and estimating the measure of infrared radiation that is produced and reflected by articles or individuals to outwardly render temperature. A thermal camera utilizes a gadget known as a microbolometer to get this vitality simply outside the scope of obvious light and venture it back to the watcher as an unmistakably characterized picture.
Some popular Face Recognition Algorithms are
o Face recognition using Tensorflow
o Deep Face Recognition with Caffe Implementation
o Node FaceNet
o Android Face Recognition with Deep Learning
Where does it is used?
Face recognition is widely used by social media platforms to attract more users in amid of competition.
Looksery App which is now owned by snap chat uses to real-time face recognition with a filter that allows the user to modify the look of the face. Similarly, Facebook has also developed a DeepFace which is a facial recognition program. It identifies humans in digital images with its nine-layer neural net with over 120 million connection weight.
Growing use of Facial recognition technology has not put an option for many companies to use this technology for identification purposes. In-fact many companies start working to provide this technology to banks and other eCommerce companies.
In Security Service:
Security is the most important aspect of this technology. Facial recognition systems can monitor people coming and going in airports. The technology is used to identify people who have overstayed their visas or may be under criminal investigation. Many countries like Australia, New Zealand, Canada, UK, Netherland, China etc. are using facial recognition technology for various security purposes like on airport, public places, on borders etc.
Besides its advantages in security purposes, we can also not deny its disadvantages. Here are the Six disadvantage of facial recognition.
There are a few concerns that Face Recognition permits governments to undermine security rights. The legislatures have no government has no limit to utilize this technology, he can use it as they see fit. Accordingly, they could keep an eye on their residents and single them out for what they say or whom they meet.
In democratic nations, laws can be made to prevent misuse of the technology. But what if rebel components inside government organizations misuse this technology to pursue their private agendas. That is the thing that got a few people concerned.
But many countries have not made any effective laws which could guarantee the prevention of misuse? These are some question which are yet to be answered.
Individual Privacy Concern
The facial recognition systems are capable to recognize the subject from millions of images and video sources. The sources include CCTV, Smartphones, Social Media and other online activities. Police Department, Intelligence Agencies using these techniques to monitor & track down criminals. But what if they decided to track anyone for personal reason,
Face recognition technology enables the mass tracking either by govt. or by any private company. It is just like taking your DNA or Finger Print without you know about it.
Data Privacy & Data Theft
Another reason of concern is data privacy. The data includes images and videos files which are stored in cloud servers which may at risk of being stolen by hackers. In such type of data breaches, the information will get into wrong hands, although the companies are seriously taking steps to prevent any possible breaches but yet no one can guarantee for 100% protection.
Lack of Regulation
Government around the world are yet to pass laws to check and limit the use of facial recognition techniques. With the increasing use of Business houses adopting facial recognition, needs a plan to avoid risk for data breaches which can only be possible by government intervention.
Read Other Stories:Bold and Beautiful ways to get through the Coronavirus lockdown.