June 22, 2023
Innovative surveillance tools used at the height of the pandemic have revealed how computer vision and machine learning can benefit public health.
Image: Project Odysseus Team
Developed and deployed in London during the COVID-19 pandemic, the monitoring tool proactively informed over 700 social distancing interventions during the height of the first wave of the pandemic. bottom.
Researchers from Cambridge and Warwick Universities, working with the Alan Turing Institute, have developed and implemented the tool and provided it to the Greater London Authority (GLA) and Transport for London (TfL) to promote social distancing. It provided near real-time data on adherence. Cityscape of London.
This has been in place for 18 months, including the first two lockdowns in 2020 and the third in 2021.
Driven by digital twins, computing and data infrastructure, and state-of-the-art machine learning algorithms, this tool has helped inform policy decisions such as the effectiveness of the 2-meter rule for COVID-19. . We quickly identified where we needed to adjust the road spacing.
These interventions include moving bus stops, widening sidewalks and closing parking lots to ensure social distancing.
In this project, known as Project Odysseus, researchers processed live feeds from over 900 traffic cameras to monitor vehicle and pedestrian activity, and algorithms calculated distances between pedestrians.
Individuals could not be identified using low-resolution footage.
This project reused air pollution tools and datasets from previous work on air quality monitoring by the Alan Turing Institute and extended this monitoring to pedestrian activity as well as vehicles.
Project Odysseus has successfully provided these activity indicators via an online interface for policy makers to use conveniently. Through a web application that acts as an ‘early warning system’, policy makers can see the impact of their decisions in near real-time in an easy-to-use format and adapt their actions accordingly.
Mark Girolami, principal investigator at the Alan Turing Institute and co-first author of the paper, said: “During the lockdown due to the pandemic, the Alan Turing Institute provides daily information reports to the TfL and GLA to assist policy development. and evaluated the effectiveness of the policy,” he said. These policies apply to social distancing in all boroughs of London.
“The report is based on an analysis of video feeds from all street cameras and relies on a number of advanced machine learning and computational statistics techniques working at scale.
“The Cambridge engineering students who joined the Turing team made important innovative contributions to the overall system implemented and ensured its efficient operation.”
He added, “Future work on this project includes validating early warning detection systems, improving the overall accuracy of the digital twin, and recommending more human-machine teaming or ‘human participation’ (policy-making (very easy for users to use) and examples of machine learning applications that are transparent and easy to ask questions. “
