Skip to main content

Behind the mask, how image tracking & data intelligence can keep us safe

data
(Image credit: Image Credit: Geralt / Pixabay)

Alessandro Chimera, Director of Digitalisation Strategy at TIBCO examines the implementation of digital technologies to track and analyze the use of face coverings, headgear safeguards and other protective equipment in public spaces.

With the need to wear a face mask or some form of protective face-covering very much a part of the normal way of life in countries all around the globe since early 2020, we can perhaps remind ourselves that this core health-tech product is an essentially non-digital innovation. Apart from a few prototype ‘smart masks’ with their corresponding smartphone apps, most face coverings are really just a piece of material and some elastic straps.

Of course, even before the pandemic, many people had become used to wearing some form of face mask, some type of helmet, hair covering or protective headgear. From the top downwards, many workers are also well-accustomed to donning protective equipment, jackets, gloves and even body armor, depending on their profession. 

But as technically simple as most of these protective coverings and shields are, we are now more concerned than ever about compliance and the need to wear these layers. This being so, could technology be a defined route to making sure we tighten all the straps that need to be fastened and keep everybody safe?

Beyond RFID, into image intelligence 

In the past, we might have considered the use of RFID tags as a workable solution for monitoring whether a piece of protective equipment was present and being worn by a person. But as smart and useful as RFID is, it has limits and its application only goes so far. Some personal protective equipment is too disposable to consider tagging and – perhaps most importantly of all – even when a bright and shiny RFID signal is correctly tracked, it can’t tell us if a person is wearing their protective coverings in the correct way.

Although it might sound strange, it is of course possible to wear a protective jacket incorrectly, perhaps unzipped, or perhaps without an outer life-vest if at sea. A worker on a construction site carrying their hard hat at their waist may have a good hold on it, but they’re not wearing it correctly. Then as we all know, an anti-Covid facemask worn on the chin with mouth and nose exposed is never correct.

So, what can we do?  

Pioneering work and prototyping use cases are already in place to use image tracking software to capture users’ behaviors and give us the insight we need into how protective equipment is being worn at work, plus also at play. 

The human-to-digital dynamic

There are now some excellent reference points and exciting pilots underway that serve to explain the mechanics of the human-to-digital dynamic and the gateway it opens up for the implementation of AI and integrated data analytics.

Areas of the transportation industry have been looking at how technologically and physically possible it is to track the use of face coverings in the post-pandemic recovery period. Appropriately anonymized, they have been examining the option to create heatmaps of passengers and so be able to track congested areas. Deeper into the analysis, they can even employ technology that will track whether a user is wearing a facemask and whether they are wearing it correctly.

Focused on people, not personal information, this system could build a bank of data results detailing the how, when, where and what of passengers are doing inside the train station, cruise ship or airport terminal, for example. Image tracking software could then be built to deliver an algorithm that can be trained to understand the difference between a beard and a mask, the difference between a mask and a scarf and the difference between whether a mask is worn correctly or incorrectly. 

This process involves taking a live video stream feed in real-time and being able to apply AI to continually train and improve the algorithm. The idea is to acquire a data stream from a selection of ceiling, wall and public space cameras in operation. IT teams can then apply trained statistical models employing the use of specially built dedicated AI algorithms to analyze the data ingested here and classify people into mask wearers, non-mask wearers and incorrect mask wearers.

Data: connect, exchange, integrate, analyze

To make all this user-level intelligence work, there are some intelligent mechanics in motion at the backend. The technology enabler here is a software and data platform that can handle data analytics, streaming and integration. In order to exchange information detailing when and where people are wearing protective equipment, the data platform needs to be able to acquire, ingest and integrate data from a variety of different sources.

Equally and correspondingly, the data platform also needs to be able to communicate outwards and report to people (or support staff working in the areas that people traverse or populate) in order to tell them if they are in breach of compliance regulations. This means the platform needs to be source-diverse in its output as well.

There’s also a Master Data Management (MDM) function here (i.e. we need to know about all use cases and all protective equipment types in all the shapes and sizes it comes in). Looking deeper, Application Programming Interface (API) technologies can be used to exchange data to an external cloud-based service. The whole technology proposition is one of data intelligence, integration and analytics.

These examples extend beyond airport passenger tracking systems and antiviral face mask-wearing. Let’s also think about working environments where safety is a real concern, such as construction building sites, oil and gas rigs, utility facilities and other major civil engineering environments, where wearing some form of professional personal protective equipment is mandatory. 

If we can build intelligent systems that alert supervisors to any breaches in compliance, then we should sustain few injuries. In return, fewer injuries potentially means lower insurance premiums, more productive workforces and more profitable business models.

Analytics streams: historical data & real-time data

Bringing all of this story’s threads and concepts together, organizations, government bodies and other public and private authorities can look to these techniques as a means of making our lives safer. An analytics dashboard can also be used to visualize important historical statistics and compare those records against real-time information. 

The real strength here lies in the ability to track real-time data against historical data where we have metrics we can use as upper limits, thresholds and performance levels that we know we need to achieve for safe operations to be ensured. 

By tracking live data against a specific point of time in the past, we can monitor progress, whether we’re tracking face mask adoption, hard hats on heads or any other environments where we want to be able to get a granular view into the way the world is operating. This isn’t a case of Big Brother watching us, it’s more a case of Big Mother looking after us. We can make the world a safer place in the future and data will be fundamental to every innovation tomorrow and in the days ahead.

Alessandro Chimera, Director of Digitalisation Strategy, TIBCO

Alessandro Chimera is Director of Digitalisation Strategy at TIBCO.