Alexander: the tool supports two rounds: filtering and rating. In the filtering round, where you can simply mark an image as 'yes'/'no' (usually done by 1–2 people from the core team), and a rating round, where the judges can rate each image from 1 to 5 (typically you'd want to leave ~300 images for this round). What we do is then have a 3rd round where we ask judges to pick the top 10 out of the top 30 top-rated images and rank them from 1 to 10. This is also the method used by the international jury last year, using this tool. Currently I don't have it set up anywhere, if this sounds good to you let me know and I'll make a copy so you can test it. In fact it's a good idea if anyone who is thinking of using this tool say so now :) last time it saw wide usage was in 2012, and since then issues might've popped up that need addressing so it would be good if I had as much time as possible to address them.