This should be in the same neck-of-the-woods of the interests of the iNaturalist community.
The crowdsourced science platform Zooniverse just released a project that asks users to identify waterfowl from aerial photos. The data will be used to develop an image recognition dataset so land managers can easily conduct population counts.
At this point, the level or expertise required is the ability to determine whether it is a “duck”, “goose”, or “crane”. I clicked through a few and didn’t see any scaups or mergansers (ducks?) or herons (cranes?), but however that works, they do give a little tutorial on how to spot the differences they are looking for in each classification.
just spent about 20 minutes trying this out, and here are some of my takeaways for people who might be interested:
the species they’re focusing on seem to be Mallard, Canada Goose, and Sandhill Crane, but you’re just asked to indicate ducks, geese, or cranes in general. if you see any species that don’t fit into that category, you just mark them as “other” and provide a guess if you have one.
it’s very similar to the early training stages of the Cornell Lab’s Merlin tool, where you draw a bounding box around any birds in the image.
over 90% of the images I got were of Mallards, with some having no birds and some having birds that were too far away to identify. I got one Canada Goose and no cranes.
seems like a cool project and one that I could easily spend way too much time on, haha.
This is addictive. It kind of feels like one of those “prove you aren’t a robot” tests, but instead of click on every picture containing a stoplight it’s click on every duck. I love it.