One of the most obvious effects of the pandemic of COVID-19 It is the rule of social distancing that has been applied to avoid contagion. Now a artificial intelligence is able to detect if we are keeping the healthy distance with other people and alert us when this does not happen.
Landing AI, a company led by Andrew Ng, one of the brilliant minds behind Google Brain and Baidú’s AI group, has designed an algorithm that warns if people respect social distancing.
This system uses a video camera, so it can be applied in workplaces and ensure that there are no more infections. The algorithm recognizes people and draws a red line between them when they are not at a safe distance, at the same time that issues an alert so they are aware.
The company shared the details of this AI a few days ago. The most important point is that It can be calibrated to fit a security camera. Customers can define the space where people move (floor) by selecting four points in a tool. After this the system will apply the adjustments to the entire perspective image.
AI analyzes your position in real time
AI uses a open source pedestrian detection network to draw a box around people. With the data obtained in the space calibration it is possible to detect the position of users and update it in real time as they move.
People who respect social distancing are visualized within a green box. Those who violate it will be locked in a red box with a line that delimits the distance between the two, as a demonstration of the tool published on video shows.
Landing AI provided an explanation of the steps it took to develop this system and said its goal is to share it with its clients and encourage others to explore new ideas to keep us safe. “Until a vaccine is available, social distancing is our best tool to help mitigate the coronavirus pandemic,” the company said.
Companies like Amazon, who recently threatened to fire those who do not maintain social distancing, could take advantage of the tool of artificial intelligence to mount it in its different distribution centers.
👇 More in Hypertextual