Andrew NG’s technology startup Landing AI, has developed an AI tool for monitoring social distancing in the workplace.
As governments all around the world continue to advise on social distancing measures due to the coronavirus outbreak, it is of crucial importance that we employ technology to help us in one way or another during the crisis. Landing AI, a company offering AI transformation for businesses has announced a new tool for monitoring social distancing in workplaces that works with any kind of video input.
The new tool takes a video as input and tries to find people or more precisely clusters of people that stand too close to each other (by using some defined distance metric). It does so by using deep neural networks and object detection algorithms. The proposed approach consists of three main steps: calibration, detection and measurement step.
In the first step called calibration, the tool provides a user-friendly interface that can be used to calibrate the camera and obtain a bird-eye view of the scene. According to Landing AI, this tool enables even non-technical people to calibrate the system in real-time. During the second step, a deep neural network (more specifically) Faster R-CNN is used to do pedestrian detection. In the last step, the distance between the detected pedestrians is computed by projecting the bounding boxes to the bird-eye view. This allows the tool to highlight people in clusters or people who are standing too close to each other.