As I wrote in my previous post here, during the third weekend of September 2020 we organized the first event for drones’ operators. The goal was to take as many as possible aerial photos so that Ewa Guard could analyze them and locate illegally dumped tires.
The response was much better than expected! Tens of drones’ operators across Poland joined the initiative and literally within the first hours generated hundreds of gigabytes of data and lots of images. THANK YOU!
Out of sheer curiosity, we of course briefly looked at the pictures and sadly said: “there are no tires!”. UPDATE 2020–09–24: then we asked volunteers to help us analyze the pictures manually. Very high resolution of images in high amounts got the teams very tired quickly. Needless to say that after all there was nothing exciting in browsing through images just to find maybe some that accidentally included small tires. Nevertheless, the teams were able to find a few of such images.
Then we sent all of the pictures to Ewa Guard for fast and AI-assisted analysis. We were quite surprised seeing how many tires we actually found over the weekend! I have uploaded some of the results below.
Ewa Guard analyzed the images with AI and located several tires dumped in various places in Poland.
I am super excited as the engagement of literally everyone who joined the initiative over the weekend and the community we’re building around it helped us make so many things possible in terms of future developments of Ewa Guard. First of all, it is now possible to start building the map of the clean world. Secondly, we got so many images that it is now possible to train Ewa Guard to detect dumped plastic bags and other types of trash as well. Also, Ewa Guard is now updating its AI models to increase the accuracy of detections. Thank you once again. We are now thinking of organizing such events on a regular basis. Plus, we are working on a few cool ideas as well. I will keep you posted. Follow us on social to stay up to date.
Here are some example images Ewa Guard analyzed this time: