My research group actually has a project in beta at the moment on there: https://www.zooniverse.org/projects/rossmcwhirter/astro-ecol...
We're asking people to tag animals in infrared images. There isn't much data on there yet while we figure out the simplest way to get people involved (feedback driven). The site engine is pretty good, it's very easy to set up your own data campaign. It's quite challenging to frame your problem in a way that people with no domain knowledge can label your data accurately.
The only disadvantage (from a research perspective) is that you need exposure and traction to really succeed, because you need a huge number of classifications (multiple times per image) to get good results. If you look, there are a lot of projects with quite a lot of classifications but 0% completion. Certain categories are just popular - e.g. exoplanet detection and astronomy always gets a lot of interest.
Have you used Zooniverse in the past? If so, what was the outcome/experience like?