Face Recognition Technology: Recent Developments


It seems that machines are becoming more like super humans. They could be trained on huge set of data gaining experience much faster than a human does. Machines have cognitive intelligence, but how about emotional intelligence. Apparently, most of us feel that they are just dumb machines. However, now they could also understand emotions and act according to them. Machines could understand the emotion of the user through analyzing enormous data sets available digitally such as social platforms and usability traits of the person. Surprisingly they can now also understand the facial expression of the person. Let’s go through some of the technology that makes machines enter our lives deeply through face recognition.

Face Recognition

Affectiva is a leading AI company. It measures the emotions of the users in their context using a simple webcam. The software analyzes facial expressions such as smile, smirk, frown and furrows which give indications for surprise, enjoyment or anger. This information might seem not important; however, it paves the way to develop an enormous number of applications. It gives the machine the so called "emotional intelligence". The idea was used by many companies as it allows them to know more about their customers. 

The software was used to analyze users’ emotions while watching ads. It could give the companies detailed report of which second surprised the customer or made them angry. This tech can give very precise opinion of what is being shown. The company gives this analysis in different demographic areas and to the variety of content shown. Such information would help companies customize their product and their marketing strategies much more effectively. Moreover, such application can be used to evaluate people response in political polls during debates. The technology would support politicians using it greatly. Statistical analysis of thousands of people will show their emotions while the politician is talking in real time. Accordingly, the message being delivered could be optimized easily. There are also game developers planning to use emotional intelligence to offer more user oriented gaming experience. Opponents of this technology assume that this will give excessive control over people much more than what is available already online through social media and browsing traits. Companies might use the technology without user's consent or awareness. Rana El Kaliouby, CEO of Affectiva, said they are committed to inform people before doing anything and having them signing a consent form. Opponents are skeptical about this commitment and say it would be neglected by time.

Compare prices of mobile phones and tablets here

It seems there is even more that can be achieved by facial recognition. Affectiva can measure heart rate through face reflection from a webcam. This can have great implication on health and lifestyle. People doing sports can monitor their heart rate easily without having to wear any sensor. Another very helpful application is analyzing emotions and face reactions while driving. The system can recognize when you are active and when you might fall asleep. It may give an alert to the driver or it may even stop the car at the side of the road. Such applications can help avoid many accidents that happen due to drowsy driving or any sort of distraction. Face recognition may also be used to get some inferences about people. Machines are used in hiring to predict good candidates. However, machines are not free from being biased. Joy Buolamwini, MIT student, was working on face recognition. The algorithm she used was not able to recognize her face. She proved that developers of the algorithm did not train the algorithm to include all the wide range of skin tones and facial structures. This algorithm was tested in November 2016. However, algorithms developed a lot since that time. There is also much larger dataset available to train the machine. Nevertheless, it is still too risky to use this young technology for critical purposes. Face recognition is being used in smartphones to process and add filters and effects to video calls in real time which is available in Snapchat and Facebook messenger. New smartphones are using face recognition as a credential to validate the identity of the user and unlock the phone. Companies are competing to reach a unique solution in face recognition that could not be easily tricked. For example, someone may put a photo of the person in front of the camera to pass the lock. 

Samsung developed a new technology by making iris scanning discover patterns that are unique to every person representing a unique biometric like fingerprints. The eye is illuminated by infrared diode and then this light wavelength is captured by a special camera. Normal front camera is not able to detect infrared light. The image is then stored locally on the device. No cloud is used to better secure this method. Although, this method provided better security, surprisingly a high quality imitative eye could trick this system. Accordingly, this will remain unsuitable for critical and highly confidential data but it will remain suitable for the public. ZTE developed a new way to unlock smartphones. It does not use infrared camera like Samsung, but it rather uses the mobile high-resolution front camera to scan the eye. The software recognises unique blood vessels for each eye. In addition to security, this technology enables eye tracking to control suitable applications with eye movement such as scrolling up and down or right and left.

Compare prices of mobile phones and tablets here

Apple Face ID developed a more secure system. The system depends on an array of sensors that are on the top of the phone. The face is illuminated by infrared flood that works even without light. Then 30,000 point matrix is formed capturing every detail of the face while the user is moving their head into circular motion as it is asked by the system the first time to identify the user's face. It is much more secure than regular face recognition as the depth of the face is identified. This system is very hard to be deceived as the deceiver needs to have a very accurate model of the user. Apple said the error probability is only 1 in million. The software is capable of capturing tiny movements of the head. This enabled apple to make interesting applications such as Animoji. The application captures exact face motion. The motion can then be shown in real time in a chosen animation character.