By: Cyberlink

The Internet of Things (IoT) was quick to take the world by storm and inch its way into much of daily life through internet connected devices that make cities, homes and workplaces smarter. When IoT is paired with Artificial Intelligence (AI) technology, you get AIoT (Artificial Intelligence of Things). This is when AI is embedded into the infrastructure of an IoT device, such as its chipset, unleashing the power of data and machine learning. AIoT provides greater value by enabling connected devices to not only perform desired tasks, but to also constantly improve through machine learning. This results in smarter, more efficient operations and greatly enhanced user experiences.

The partnership between machine learning and IoT to create AIoT has paved the way to many innovations for a more connected future. When combined to AIoT, Facial recognition makes a compelling use case for businesses and consumers across industries.

AIoT Common Use Cases
AIoT is a transformational technology. We cannot overstate just how rewarding it can be for many businesses. Let’s look at a few of the most common applications for AIoT today. However, there is much more potential to be realized.

Access Control
One widely used use cases for AIoT is for access control. This could be through a variety of outlets and technologies, including smart locks on doors or cabinets, and mobile device and equipment login. With a smart lock and door access, AI technology will help the lock remember specific rules. For example, it would only allow permitted people inside, through a code or biometric technology like facial recognition, as well as follow a rule for only unlocking during working hours and during the week.

Smart Signs
Another strong use case for AIoT is for digital signage and interactive kiosks. The latter is experiencing increased adoption due to the Covid-19 pandemic. For example, a kiosk placed at the entrance can ask a user a set of health questions to make sure they are feeling well before being allowed inside a store or restaurant. Depending on what the user responds with, the AIoT device will prompt the next response and appropriate action.

Improve Security
AIoT-based devices have flooded both residential and commercial markets in recent years to enhance security and protection. Whether offered through complete service packages by providers such as ADT or as devices that can be installed by end users and connected to their wifi, these solutions already integrate AI-based features such as motion or intruder alerts, remotely programmable access codes, and more. Within a few hours, a home or facility can be secured and offer powerful at-home and remote monitoring features.

Vision Technologies and AIoT
In the AIoT use cases mentioned above, there is a clear benefit to using a vision technology to perform identification, authentication and access control tasks. Vision technologies are also referred to as biometric technologies. They are generally applied through either facial recognition, fingerprint or iris reading and authentication.

Iris recognition works by measuring unique patterns in an individual’s iris to verify identity. Facial recognition technology works by identifying facial vectors and features and matching them to an individual. Fingerprint recognition works through a specific sensor that is designed to pick up on fingerprint patterns and then match those to an individual.

A key concern with iris recognition is that there are only select and specific cameras designed to perform these tasks. They are also expensive. Compared to facial recognition, there are more cameras on the market that can be tooled to perform facial identification than iris recognition. With fingerprint recognition, on the other hand, hygiene can cause issues. Fingers can be dirty or oily and therefore not recognized by sensors. Dirt can also damage sensors.

Of these three methods, facial recognition technology is recognized to be superior because it is more precise, flexible, affordable and hygienic. As such, we will be exploring how facial recognition can both enhance current and enable new AIoT use cases for the remainder of the article.

Integrating Facial Recognition into AIoT Enhances Existing Use Cases
Facial recognition can enhance the value of AIoT devices and use cases, greatly increasing their adoption. Let’s look at a few examples.

Access Control and Identity Verification
While we have previously touched on AIoT for access control, it is worth detailing exactly how facial recognition technology can be involved to enhance these use cases.

Facial recognition can transform access control and monitoring systems for homes, residential complexes or commercial facilities. The technology enables precise, instantaneous identification of individuals in front of a camera, enabling contactless access of authorized people or sending instant alerts if block listed people or intruders are detected.

Integrating facial recognition to employee time-clock and access devices or systems can streamline the process, reduce errors, and eliminate risks of employees sharing access cards, while monitoring for unauthorized entry attempts.

Access and Identity verification doesn’t stop at home and facility protection. Imagine a manufacturing warehouse. Much of the equipment can only be used by designated personnel. While some machines might require a physical key or numerical code, both can be stolen and lost. If the machine is equipped with facial recognition technology to grant access to designated personnel, there is less risk and more security and control. In addition, if a warehouse manager wants to set rules so that a machine can only be operated during working hours, an AIoT device can be programmed with those specific conditions.

Personalized Customer Experiences
Facial recognition can be used to enhance customer experiences in a number of ways.

A good use case for this is retail. When a retailer installs facial recognition AIoT devices throughout a brick-and-mortar store, they can be programmed to recognize an opt-in VIP customer and alert staff upon their arrival for a personal greeting.

AIoT devices can also be leveraged for data analysis. In the same retail use case, AIoT devices with facial recognition can capture customer behavior and demographic data. They can determine if customers express confused looks on their faces in certain aisles, or if they are more inclined to smile when they walk by a mirror. These patterns are captured by the facial recognition technology to be analyzed by AI. The retailer can then take action and rearrange their store to produce more positive experiences.

AIoT to Validate Identity and to Prevent Financial Fraud
A cyber security use case that is quickly growing in popularity is electronic Know Your Customer (eKYC) authentication. Facial recognition provides by far the most accurate and convenient technology to verify someone’s identity and provide second factor authentication, in opening bank accounts, applying for credit, conducting ATM transactions or mobile banking, buying insurance services, and using secure remote customer service. While offering very compelling validation, the process is as simple as matching the live face capture to a valid ID that is either scanned in the process or already on file.

Facial Recognition Enables New AIoT Use Cases that Wouldn’t Be Possible Otherwise
Due to the ongoing innovation happening with facial recognition and related hardware, there is immense potential for facial recognition and AIoT. This is especially true for edge-based facial recognition solutions, like FaceMe^®.

Edge-based facial recognition is when the technology is embedded in the IoT device without a need to perform cloud processing. This includes smart locks, mobile phones, point-of-sale (POS) systems, interactive kiosks, digital signage and more. Edge-based IoT devices run facial recognition within milliseconds and with extreme precision. On top of dramatically speeding up the process, the absence of cloud processing addresses data security issues and results in huge cost savings.

The best edge-based facial recognition AIoT solutions are highly accurate, can process images and camera feeds quickly, encrypt data for security purposes and work across hardware, platforms and programs. With a clear understanding of the benefits and potential of edge-based AI, hardware makers have been quick to redesign their technologies for these solutions, resulting in a race to develop more powerful AI chips and reduce cost. Some of the top and reputable manufacturers include:

• Chipsets: Intel OpenVINO, Intel Movidius, NVIDIA Jetson, Qualcomm Snapdragon, Broadcom Raspberry, MediaTek i350, among others.
• Low-cost Industrial PCs and mini PCs: Advantech with FaceView, SuperMicro workstations, Brightsign devices, among others.
• Low-cost cameras: Logitech USB cameras (for basic use cases), VIVOTEK IP cams, elo cams for temperature reading, as well as others.

Some of the most interesting new applications for facial recognition and AIoT have been enabled by these increasingly powerful yet affordable edge-based technologies.

Health Kiosks & Mask Detection
Ever since the pandemic hit, the health and safety of individuals in public and private spaces has been a priority. Mask wearing has also been normalized. In many places it is even required to grant entry. This makes such a compelling use case for AIoT and facial recognition to work together for health kiosks. The kiosks, when equipped with facial recognition, use a camera to detect if the individual is wearing a mask, and if they are wearing it properly covering the entire nose and mouth. They can use a camera with thermal scanning capabilities to read temperature. This ensures no one enters the property with a high fever. This could be granting access or sending an alert to a designated person to handle any necessary action.

Interactive Customer Experiences
We have discussed previously how facial recognition and AIoT can bring about more personalized customer experiences. Now, we will discuss how these solutions can also be interactive, and even more rewarding for the consumer.

Many retailers are challenged to keep customers engaged and enjoying their time in store. Digital signs are increasingly popular in malls and retail stores, providing rich-media content that can be refreshed through content management systems. It is now possible to embed facial recognition in signage at a low cost, enabling dynamic display of content based on factors such as the gender, age and mood of an individual looking at the sign. Even better, people opted in (to a loyalty program for example) can see fully personalized content based on their previous purchase patterns and other collected data.

Add product selection and payment capabilities to turn a digital sign into an interactive self-service kiosk. An opted-in customers can be automatically identified by their face, getting a perfectly tailored shopping experience, including special offers and even virtual try-on where relevant. They can even use their face to complete their purchase and perform a true contactless payment.

Smart Medicine Cabinets
AIoT is a key driver for healthcare innovation and smart medicine cabinets are the way of the future. When equipped with facial recognition for access control, connected-medicine cabinets offer greater security and control over protected substances, on top of quick, contactless and hygienic use authentication. Facial recognition adds substantial value to this AIoT use cases, on top of other typical security measures such as synchronizing with staff work schedules to only authorize access when permitted to designated personnel.

FaceMe^® perfectly exemplifies a good edge-based facial recognition solution that can help with these use cases and many more. It’s easy to integrate across a wide range of devices, offering one of the market’s most comprehensive chipsets and OS support. Its highly accurate AI engine is ranked one of the best in the NIST Face Recognition Vendor Test (FRVT). FaceMe^® can be deployed across a wide range of scenarios, including security, access control, public safety, smart banking, smart retail, smart city and home protection.

The Future of AIoT and Vision Technology
Facial recognition might well be the top enabler to the future of AIoT technology. It makes AIoT solutions safer, smarter and more human.

However, there are still remaining barriers to reap its full potential. Some of these barriers are physical, such as hardware or inherent limitations. Others are social, including concerns over privacy and data security.

First, when considering the physical limitations, it’s important to remember that there is no one size fits all approach. Businesses all have varying needs, as well as budgets, and should look to solutions that will work for them. The best facial recognition solutions can be scaled up or down to suit the specific needs of a business and its use case.

It’s also important to consider accuracy and controlling environments to be more conducive and inviting to facial recognition technology. This means paying attention to factors that can hurt accuracy, such as lighting, camera position, and lens cleanliness.

On social barriers, facial recognition has come under scrutiny and is critiqued for surveillance uses and biases. This does not mean the technology should be abandoned. Rather, there is work to be done on explaining facial recognition’s many benefits such as security, convenience and delivery of new, better experiences. There are also needs to educate both public and private entities on ethical use, as well as proper regulations to be put in place.

There is so much more good facial recognition can and will do for our world. Interest in the technology is growing. Many technology suppliers are even saying the coronavirus pandemic represented a pivotal moment for this type of biometrics technology and its deployment through AIoT.

We’re excited for the potential of facial recognition. We’re committed to innovation; to delivering solutions that businesses need and consumers are comfortable with, and to contributing to create safe touchless environments and amazing new user experiences.

For a full overview of facial recognition, how it works and how it can be deployed, read Edge-based Facial Recognition – The Ultimate Guide.