Venturing into realms that I am less familiar with here, but it seems to me that if you are making your own conclusions from the photo (and not really using any of the other observation data per se) it should be fine, even with the license.
The observation license seems to cover things like the location info, date, other tags the observer makes, description, etc. So if you’re not using that information, I don’t think it would be an issue (probably, I’m definitely not an authority).
But I think it’s key to emphasize that this is a really gray area without a clear answer that I am aware of. I would suggest checking out the beginning of this thread if you haven’t seen it before.
A little way down @kueda notes “I should also point out that while photos are subject to copyright in most jurisdictions, observations may not be, as they represent facts about the world and not necessarily the kinds of creative works copyright was designed to protect (unless you write a description). On iNat we assume observations are copyrightable, but the only way to really test this is in court.”
So I agree that it’s likely that that observation data wouldn’t be eligible for copyright, BUT there isn’t a definitive answer on this given the complexity of the law (potentially in many different countries) right now. I do think an important point is respecting the wishes of users however. In that thread, the OP discusses how iRecord harvesting their observations (when they had explicitly set their license to prevent others using the OP’s info) was discouraging them from participating in iNat further. So I think the negative consequences of using people’s data when they’ve said they don’t want others to do so via their license choice are real.