Real-time High-resolution Measure of Social Distancing

By now, everyone knows about the need to "flatten the curve," i.e. slow COVID-19 enough to spread out healthcare utilisation over a longer timescale and prevent a catastrophic capacity overrun. We realise that our best hope to flatten the curve is to engage in social distancing, but how effective are our efforts, how long will we need them, and will we be able to gradually transition back to normal life while maintaining safe levels distancing?

To help answer these questions, dRISK.ai have developed a real-time measure of the effectiveness of social distancing based on directly observable, publicly available, but totally anonymised CCTV data. Using specialised object detection technology on London's huge collection of public CCTVs, our system can determine the level overall human density, its hour-by-hour change, any hotspots of activity within London, and how well people are doing at keeping the mandated 2m separation from one another.

This system allows

* epidemiologists to better predict the timecourse of the virus,

* policymakers to fine-tune guidelines for social distancing and travel directives to better guide the course of public response, and

* the public to directly visualise the general state of social distancing and understand the need to abide by issued policies.

Because of our preexisting mandate to characterize transportation risks, dRISK have been compiling data from these cameras since February and monitoring the statistics of distancing behavior. Despite the dire situation imposed on us all by COVID-19, our team has felt a sense of hope and increased resolve as we have seen Londoners adapt, learning to change their behavior in public to keep the 2m separation in a great number of cases. We now hope to help aid the transition back to a growing economy and thriving city, all whilst still keeping ourselves safe with measurable adherence to social distancing norms.