After about a month, this weekend we passed the 100,000 craters marked point, so it seemed like a good time to give you an updated picture of what, well, the picture looks like.
The images that you have been doing have not quite been randomly selected — we’ve put emphasis on a specific image strip which happens to also be the one that I did manually (identifying around 11,000 craters on). The reason for this is that we need to prove that Moon Mappers works and can yield composite results comparable to crater experts (this is what we hope to show at the conference next month).
This particular image strip was divided into 185 450×450 sub-images. At this point in time, an average of 3.14 users have looked at each sub-image, while up to 7 users have looked at 2, 6 users have looked at 7, and 22 users have still only looked at 1 sub-image. The goal is to get up to an average of at least 10 users per sub-image.
But, we can still look at the results (click to make bigger!):
In this image, all the pale blue markings are the result of hard work done by you. The red markings are the result of a clustering code that groups markings based on crater location and size. Only craters that were marked by at least two people have been output at this time (which actually cut the results in half).
If you look towards the lower-middle of the image, you can see that this is a good thing because there were lots of stray markings around the two larger craters. The clustering code couldn’t find another matching crater, so they were rejected.
On the other hand, if you look below this, just around the legend, you’ll see some pretty clear craters that were only marked by one person; even though they’re good craters, since only one person marked them so far, they were not output from the clustering code.
And this is why we are going to keep the images in use until we have an average of about 10 people looking at each one.