TMCnet Feature
June 16, 2020

The Ultimate Guide to Facial Recognition Technology

For many, facial recognition technology has been associated with an exclusive group of institutions, or even with science fiction movies. However, thanks to rapid advances in the field, this technology has real business value for companies of any size.

In its essence, facial recognition is a part of applied machine learning and is used to identify human faces with the help of computers. This can be applied to many real-life business challenges in many different industries.

How does facial recognition work

Facial recognition systems rely on geometry to “see” a human face. Things like the distance between eyes or forehead and chin are measured. These distances are often referred to as facial landmarks, which make a face unique. Put together, the facial landmarks make up a facial signature, which is nothing more than a mathematical formula that represents one specific face.

The purpose of creating a facial signature is to match it to photos in a database and identify who that person is. In other words, the facial recognition systems need to have an already identified picture of a person to be able to recognize her in the future.

This is how Facebook (News - Alert) knows to suggest the right people to tag in a group photo, or how security agencies around the world can detect wanted criminals in a crowded space.

How facial recognition can be used

Facial recognition has solid business applications in many different settings.

Simplify authentications and minimize fraud

Facial recognition technology can be used in place of passwords, which is something Mastercard has done with their Identity Check. It’s a payment application that allows consumers to approve payments with a selfie. The app also asks the users to blink when approving payments to make sure that pictures are not used to make fraudulent payments.

The overall concept of how Mastercard has used this technology can be applied to any area where a transaction needs to be verified. This can be a purchase, a login, or approval of terms and conditions. It minimizes the friction that is otherwise present in these processes and can make it easier for customers to use certain products or services.

Detect information not visible to the human eye

Contrary to popular belief, humans are very bad at reading each other’s facial expressions. The reason lies, perhaps, in the existence of microexpressions, which largely constitute how humans convey their emotions.

Pavel Akapian, Machine Learning and Computer Vision Team Lead at InData Labs, says:  

Face recognition and face detection applications based on deep learning help understand the interpretation of the human face. A microexpression is a very brief, involuntary facial expression humans make when experiencing an emotion. They usually last as little as half a second and they can be hard to catch. This can make it difficult to gather certain information. That’s where the technology comes in handy.

Pain measurement in healthcare is a great example of that. Accurate pain assessment is difficult for many different reasons. Firstly, pain is a subjective feeling, which makes it difficult to make an accurate assessment based on the patient’s description alone.

Secondly, there are biases that might prevent healthcare professionals from accurately assessing the pain lever of a patient. For example, women’s pain complaints are often scored lower than men’s. The same applies to people of color.

Lastly, the patient might not be able to convey their pain level at all because they are unconscious, or have difficulty communicating. With facial recognition, one can detect facial expression nuances related to pain, thereby bypassing communication and bias hurdles. 

This same logic can be applied to any procedure where facial expressions are vital for an accurate output, and facial recognition can take a lot of the pressure off users to provide an accurate examination.

Enhance the customer experience

One of the most experimental and perhaps the most exciting, applications of facial recognition can be found in marketing. Facial recognition is slowly being used in stores around the world in ways that extend beyond preventing shoplifting.

While retailers have become very good at collecting information about the online purchasing habits of their clients, the same cannot be said about the brick-and-mortar stores. Given proper consent, facial recognition can significantly improve customer segmentation and thereby enhance the overall customer experience.

Combined with online shopping data, retailers can get a deeper understanding of their customers’ shopping habits, and offer a more personalized shopping experience.

For example, by recognizing a returning customer when he walks into the store, a retailer can offer everything from special discounts, product recommendations, or shopping assistance in-store to the person.

It’s also possible to integrate facial recognition systems with other software a retailer might be using, such as CRM, loyalty clubs, point of sales system, etc. This can be especially beneficial for luxury retailers where excellent customer service is often a big part of a brand’s offering.

What to consider when using facial recognition

Despite its many benefits, facial recognition technology isn’t perfect. In particular, there are three main areas that businesses should be aware of when they implement this technology.

Flawed results

Facial recognition technology isn’t always good at accurately recognizing faces. Errors often accrue when pictures are taken in poor lighting or with a device that produces low image quality. This can be fixed by investing in better equipment for capturing the images. On the other hand, inaccurate results also occur when the database of pictures lacks information.

Privacy concerns

While some praise facial recognition, others question what implication it has for the privacy of personal data.

Consumers are concerned about being tracked without their consent. And even when it’s done with their consent, they are worried about data breaches that have, unfortunately, become more and more common.

While businesses can’t combat these concerns completely, it is possible to offset them significantly by investing in cybersecurity and being thorough about consent for information gathering. It’s also important to be open to how the data is used and show the customers that their data isn’t being misused for malicious purposes.

» More TMCnet Feature Articles


» More TMCnet Feature Articles