Proving the Business Case for the Internet of Things

Omron adapts fisheye camera for automatic people detecting

Steve Rogerson
November 20, 2018

Omron has added its own algorithm to a normal fisheye camera to let it detect people from above without using thermal sensors or relying on movement. The Japanese company was demonstrating the device, due out next year, at last week’s Electronica trade show in Munich.
“It is a fisheye camera to give a big view below the camera,” explained Gabriele Fulco (pictured), European product marketing manager. “To this, we added an algorithm that can recognise the shape of the shoulders and head from the top.”
This is for use in applications such as building automation, for controlling door access, and in retail. It can count the number of people and track them through the building.
For similar applications, the company has a thermal sensor that can also detect people who are standing or sitting. The plan next year is to expand the array size from the current 4x4 to 16x16 or maybe 32x32.
“The bigger the array, the better the resolution of what you are seeing,” said Fulco. “With bigger arrays, you can see more clearly the shape of what you are looking at.”
Also on thermal sensors, the company is planning a version next year that will detect up to +200°C compared with the normal +85°C.
“This will be for cooking applications in kitchens and some industrial applications,” said Fulco. “We are also planning a version with Modbus for refrigeration applications. We want to expand the product range more and more.”
Another demonstration on the stand had a seismic sensor. The idea is it can be used in an industrial setting to shut off machinery automatically if it detects an earthquake. This uses a three-axis accelerometer with an Omron algorithm that has been patented worldwide.
“It can discriminate between the vibration of an earthquake to other vibrations, such as say a truck passing by,” said Fulco.
For shopping malls, the company has introduced digital signage that can detect information about viewers, such age, gender, mood and so on. This allows content to be shown that matches the viewer. It can also be used for face recognition systems for checking people in and out of buildings.