Situational awareness in uncertain times
As retailers and property owners adapt to living with the presence of Covid-19, one artificial intelligence company is offering new ways to help them keep their people safe.
Hamilton-based Aware Group has AI at its core, offering everything from advanced analytics to the Internet of Things (IoT) development and automation services.
But the company which started just over four years ago is more lately winning plaudits for its new platform, Populace, which focuses on counting people.
This area of business took shape for Aware Group when several organisations in the education sector approached them with the same need in mind.
“They wanted to solve some problems around optimising the usage of a facility,” says Travis Corrie, Head of Commercial at Aware Group.
Figuring out exactly what areas of campus students were using was an inexact science.
“They were asking, how can we replace all the manual reporting that needs to get done with a clever piece of technology?”
Combining camera feeds with AI
The answer lay in combining an existing resource – CCTV camera footage, with AI algorithms designed to count people and spot patterns in crowd movements. Aware Group has expanded the concept to allow retailers to get a better handle on where people are in their stores, how long they need to queue for service and to identify where poor store layout may impact on a shopper’s experience.
“The space between people is a hot topic at the moment,” says Corrie.
During the lockdown, physical distancing to limit the spread of the virus was employed by retailers, who taped separation lines on the floor and asked shoppers to form spaced-out queues at the front door, letting a handful of people in at a time.
How effective were those efforts? Aware Group’s AI system can tell you exactly how people comply with physical distancing rules through analysing those CCTV cameras in real-time, allowing store managers to address choke points that might be putting shoppers’ health at risk. It’s unfortunately a provision we may need to draw on again in future if new outbreaks of Covid-19 emerge.
“We can also determine ‘dwell time’ in a particular area, how long your customers look at a specific product shelf within the store. Queue counting is possible too,” says Corrie.
All of those functions offer important intel for retailers and can be delivered as long as Aware Group can access those camera feeds. The system can be totally cloud-based, so doesn’t require organisations to invest in any dedicated hardware.
The Populace solution runs on Microsoft’s Azure cloud platform, where the video data is analysed, allowing customers to easily scale up the processing of footage as more stores or properties are covered.
Facial recognition not ready for prime time
The software is clever enough to detect facial expressions, which can help determine the sentiment of shoppers or students. Is someone frowning because they are having to queue too long to pay for their groceries? Is a student frustrated because there’s too much background noise breaking their attention? Populace can determine that, delivering reporting and visualisation of the data to help inform decision making.
But Populace draws the line at identifying people individually from those CCTV feeds.
“I don’t think people are quite ready for facial recognition yet,” says Corrie.
While many of us use facial recognition to unlock our phones and authenticate our identity for web applications, he says trust in the technology for broader uses, such as identifying individual shoppers in a store and tracking their movements, isn’t there yet.
Aware Group uses the video footage for the specific purposes of tracking people’s movements without identifying them.
“As soon as that is done, the image that’s captured is then either blurred out or deleted,” says Corrie.
“And there are ways of keeping that data on-site as well. if it’s, say, a university, you can do all the computing on-site, on-premise, and none of it would leave.”
Aware Group also sees major potential for Populace as 5G mobile networks enable IoT sensor networks to deliver more real-time information that can be analysed in the cloud or with edge computing devices.
“It’s being able to deploy easily and being able to scale up and down rapidly,” says Corrie, who is also exploring the use of the Populace platform to help local government bodies monitor pedestrian and traffic flows around our cities and towns.
“One of the major benefits is having that low latency and high bandwidth that comes with a 5G connection.”
Separate to Populace, AI is also being harnessed by Aware Group for other uses, such as audio classification. A recent project saw Aware Group work with the Port of Otago to monitor noise at its busy container terminal at Port Chalmers.
The Port is required to report on any noise level breaches due to its operations and had been doing so based on anecdotal reports from people and decibel readings. With Aware Group onboard, the Port installed microphones around the terminal to capture noise levels automatically.
“Was it a container being dropped? Or was it the train going by and that’s actually not in their jurisdiction?,” says Corrie.
Aware Group’s system can tell exactly, classifying the noise and eliminating bias that may see workers or members of the public incorrectly attribute noise breaches.