Measuring the Social Impact of the
Voxel51 is tracking the impact of the coronavirus global pandemic on social behavior, using a metric we developed called the Voxel51 Physical Distancing Index (PDI). The PDI helps people understand how the coronavirus is changing human activity in real-time around the world. Using our cutting-edge computer vision models and live video streams from some of the most visited streets in the world, the PDI captures the average amount of human activity and social distancing behaviors in major cities over time.
Unprecedented Impact on Social Behavior
The spread of the coronavirus has caused policymakers in the U.S. and around the globe to implement strict physical distancing orders to slow the rate of infection and thus flatten the growth curve.
In an effort to bring information and awareness surrounding the impact of this pandemic, Voxel51 is using our AI-powered video understanding capabilities to gather and analyze video data from public street cams around the world. Our team has developed the Physical Distancing Index (PDI) to measure the impact on social behavior in public spaces. This measure is non-invasive and does not use other data sources like mobile phone signals, which allow only approximate location estimates; instead we focus on specific centralized locations of interest in each city and literally watch what is happening.
The data in the graphs speaks for itself: coronavirus and the necessary preventative measures across the globe have had an intense impact on daily life. All of the above cities have seen a sharp decline in public social activity during the month of March.
What is the Physical Distancing Index?
computer vision and state-of-the-art deep learning models are able to detect and identify pedestrians, vehicles, and other human-centric objects in the frames of each live street cam video stream in real-time. Using images sampled from each video stream every 15 minutes, we compute the Physical Distancing Index or PDI, an aggregate statistical measure that captures the average density of human activity within view of the camera over time. Outputs of the detections, or positive hits, in the video streams are represented in the data points on the graph above. Note that PDI is a privacy-preserving measure that does not extract any identifying information about the individuals in the video.
Comparing the Response
With the far-reaching impact of the coronavirus around the world, we were interested to compare data from each city to see the differences in PDI relative to each city over time. But because the geographical size and the rate of human activity is vastly different in each city (or street cam view), we must account for these differences. As such, we needed to normalize the data, or create a common starting point in order to make a fair comparison so that we could examine the differences over time. To normalize the PDIs, we set the maximum value for each location to 100%, and scaled the other values accordingly. The comparison chart below plots the normalized PDIs.
Consider the Times Square and Seaside Heights feeds; the Times Square area is physically much larger and generally occupied by more people year round, whereas the Seaside Heights location is sparsely populated in the winter but densely in the summer.
What insights can we derive from this comparison? Clearly all locations we are monitoring showed a similar trend corresponding with the spread of the virus and local statutes limiting movement. The Prague response, for example, was the earliest significant drop (March 8th). The Seaside Heights feed was steadily trending downward until the recent weekend with good weather but has again fallen off at the start of the work-week.
Comparison of PDI Across Locations
Physical Distancing versus Social Distancing
We named the Voxel51 Physical Distancing Index (PDI) in response to the World Health Organization’s recommendation (ref: Washington Post
) to do so. Physical distancing is intended to reduce the spread of the virus between individuals, especially in situations where an individual is carrying the virus but does not show symptoms. We wholeheartedly believe that it is intensely important to maintain social connections during this time of physical separation.