The last few months have been quite a shock to the world as COVID-19 has spread around the world at an unprecedented rate. With growing fears about a potentially even more disastrous "second-wave" hitting the globe, our team decided that we should try and find a novel way to help people flatten the curve.
What it does
One of the best ways to fight the spread of COVID is through social distancing. But social distancing can be a challenge. Lots of people go on walks and runs everyday, and end up getting closer than the recommended six feet of distance from other people. Toronto has resorted to increasing the width of their sidewalks, but it's an expensive solution and doesn't completely eliminate the problem. Our webapp aims to solve this problem by providing the public with real-time pedestrian traffic information, so that people can time their walks and choose routes in ways that minimize the amount of close contact.
How we built it
Our app works by having drones scan public roads and paths to detect the amount of people present at a given time. The detection is done through a PyTorch CNN model trained to detect humans from an aerial point of view. The data from the CNN is piped to a backend server which maintains running averages of the amount of people at every location reported by the drone. This data is then sent to our frontend app which lets us generate heatmaps of pedestrian traffic. Furthermore, we also utilized the Here API and Google Maps API to enable users to get directions that minimize the number of people they could run into.
Challenges we ran into
Linking together the video streams from the drones with our ML model, with our backend database, and with our frontend webapp took a lot of planning and teamwork.
Accomplishments that we're proud of
We're really proud that we managed to link all four of these technologies together in such a short duration.
What's next for CovidMaps
At the moment, we've only used a single drone for testing. Obtaining more drones would be a next step for CovidMaps.