Here is a quick update about the crater mapping crowdsourcing trial I released a few weeks ago. Even though I only had a couple people work through the crowdsourcing powerpoint, the results were what I expected. In the powerpoint, I included a total of 13 Sentiel-1 radar images. Of these, only 10 had craters and the other 3 were false positives. Going through the results, only 40% of craters were mapped correctly (with some diameter variations between users). There were also a significant amount of false craters mapped due to lack of topographic data (e.g. cinder cone) and visible imagery (i.e. small city). Below is an example of a false positive. The upper left circular feature is not a crater.
I’m not sure the reason for some of the other false craters though, so I will have to look into that a little deeper. Another factor to consider is the large size of the images. Zooming in and going through each section has been difficult for me for sure. It’s definitely easy to miss a small crater, especially when there a lot of other bright little features in the radar images.
These were just quick stats on the results, but I am curious to see if and how the results would differ with a larger data set and more participants.