<- all articles

Edge AI for COVID-19: Counting people

To stop the spread of COVID-19, our society is in lockdown. Shops, restaurants and bars are all closed because of the risk that people could get infected when being too close to one another. Governments are urging us to stay at home.

But sooner or later the lockdown will come to an end and we eventually might leave the house once in a while for a drink, enter shopping districts, enjoy the company of friends and enter public places.

We yearn for human contact and our mental health is at stake. With the government looking for new, controlled ways to let us come out of our houses bit by bit, we might have the solution to make this happen even faster.

How on earth are we going to prevent the virus from spreading again? One of the solutions is already practised right now by a lot of security guards and volunteers. They spend days in front of shops and public places to make the count and make sure no more than x people are inside and can ensure their distance to each other.

We need to prevent shops, bars, restaurants and public places from being overcrowded. We need to map busy places, we need to count people.

Edge AI to count people

At Edgise we designed an intelligent piece of hardware to accurately and anonymously count the amount of people present in (public) places based on a small camera. An AI algorithm analyses the live video feed from the camera and counts if a person passes by without recognizing their faces or even saving the camera feed.

edgise counting algorithm in action
Edgise counting algorithm in action

Our counting solution can be placed at several shops, for example, to count customers coming in and out of the shop. Implementing the counting algorithm on a small device makes it possible to give live information to the customers and shop owners of how busy the place is and how many people are still allowed to enter.

In no time, we can easily control crowded places and reduce the risk of spreading COVID-19.

Sounds very innovative! But what about privacy?

No need to worry about that. The solution is designed following the Edge Computing approach. Everything is processed/analysed on-device. To achieve this, we designed a very powerful AI counting algorithm, but deployed it directly on the device. By letting the hardware and the software work very closely together, we eliminated the need for raw processing power in the cloud. This state-of-the-art technology enabled us to both use powerful AI techniques and ensure the privacy of the people. The device only saves the counts of people going in and out.


Our COVID-19 solution led us to the development of a general local AI counting algorithm based on simple camera technology. We took our solution one step further and productised it toward not just a counting solution but something much bigger. We call it Telly.

Telly, the virtual store agent you craved.

Telly can intelligently predict busy times in the building, inform customers, analyse trends, integrate with chatbots or even detect gender or age. Since Telly is an edge AI solution, no images are stored locally or on a cloud server, nothing is being recorded.

That’s what makes Telly a virtual store agent that gives you important insights with the highest respect to your visitor’s privacy. Find out more at www.telly.ai

Get in touch with us.

Simply schedule a call with us here to get all the information you need: the timeframe we’re looking at, the cost per device and how quickly we can upscale to your needs.

Written by

Want to know more?

Related articles