Skip to content

How Retailers Can Use Computer Vision to Adapt to Changing Times

August 11, 2020

By: Ben Skidmore
This article originally appeared in Loss Prevention Magazine.

Interview with Tom Meehan, CFI

Meehan is chief strategy officer and chief information security officer for CONTROLTEK. In his dual role, he leads the company’s solutions development strategy and retail-specific strategic initiatives, while championing information security technologies and protocols for CONTROLTEK and its partners. He is an LP expert in cyber security, retail technology, and information technology. He currently serves as Innovation Team Chair with the Loss Prevention Research Council, retail technology editor at LP Magazine, and cohost of the LPRC podcast.

Loss prevention professionals have been faced with the robust job to combat shrink, discourage theft, increase efficiency, improve performance, lower costs, support safety…and the list goes on. The coronavirus pandemic has sped up these challenges and threats exponentially, while presenting the added layer of health safety to the mix.

In the turbulence of this change, keeping up to date is a constant challenge, and digital technologies have frequently been discussed and continue to be the answer as future-proof options for retailers. One of the most widely discussed technologies is artificial intelligence (AI), and one of the forms of AI most easily applicable to the retail environment is computer vision.

What is computer vision and how does it work?
Computer vision is an emerging technology that enables retailers to harness the power of video to automate the process of identifying threats in real time, leading to quicker and better decisions. In simple terms, it is defined as a field of artificial intelligence that replicates the complexity of the human vision system to enable computers to “see” and understand the visual world. Using content from digital images, videos, and deep-learning models, computer algorithms mimic the way human vision acquires, processes, analyzes, and understands visual information to identify and classify objects.

There are several types of computer vision features used in different ways, but simply put, when a computer is supplied with images, it uses algorithms to analyze for distinctions such as shapes, colors, borders, distance between shapes, and other patterns to identify a profile of what the picture means. When these algorithms are complete, the computer will theoretically be able to use this learned data to find other images that match that profile.

How can a retailer use computer vision?
Though computer vision has a lot of potential applications that haven’t been fully discovered, for retailers the solutions already exist. Facial recognition technology is a form of computer vision that has been tested and proven in retail. It is particularly useful in helping retailers detect shoplifters and alert when known bad actors enter stores. A cloud-based computer vision platform even allows retailers to access information across multiple locations.

Computer vision technology can also provide traffic and behavior analytics by using real-time, accurate visitor counts and classification, so retailers can understand customer traffic by knowing a customer’s path through the store, where they spend time, and how much time is spent there. Powerful, deep-learning technology allows retailers to know the behaviors and demographics for optimized marketing, sales, and rewards program effectiveness.

How can computer vision help retailers with challenges presented by COVID-19?
Because of its endless potential, computer vision technology can be adapted to address current challenges, such as the pandemic. Computer vision features that have been adjusted in response to COVID-19 challenges include temperature screening, mask compliance, and occupancy verification. Thermal imaging, originally intended to detect intense heat for early indication of fire, can be used to screen temperature and detect elevated body temperature of individuals entering a facility. Mask detection to identify a person as a robbery threat can be adapted to detect a face mask for health compliance. Facial recognition can determine unique customer counts for occupancy verification allowing retailers to stay within social distancing guidelines.

What are your thoughts on the accuracy of thermal imaging?
You must ensure you have the correct camera or imager and have a clear understanding of its ability and limitations. Thermal imaging that is widely available isn’t medical grade. It simply uses the sensor to detect body temperature. Like any noncontact temperature screening, there are many variables one must consider, such as ambient temperature, abnormal body temperature related to the environment, distance, and the weather. So, yes, this technology can detect an elevated body temperature, but it’s just one way of helping to keep your customers and employees safe.

How can someone integrate computer vision technology into an existing loss prevention strategy?
Like most AI solutions, computer vision is what you make of it. Investing in computer vision solutions on a smaller scale won’t prevent you from expanding its use in the future, and its future-proof design means you can integrate computer vision as your retail loss prevention methods change.

CONTROLTEK’s solution CMatch AI is scalable with the ability to operate as a standalone, plug-and-play device or as a cloud solution to save information for enterprise-level monitoring. The automation of CMatch AI eliminates the need for human interaction to support public health and safety, reducing labor costs and providing real-time information. CMatch AI helps retailers reopen stores safely and streamline compliance with new COVID-19 policies, while remaining adaptable for what changes may come.


To subscribe to the printed version of the magazine and enjoy other great content, visit losspreventionmedia.com.