The computer vision system has seemingly gotten a lot better since it came out already.
My understanding is that the basic problem is that the computer vision system was designed without two abilities that will likely be necessary to reach practical functionality:
- it doesn’t consider location at all
- it only trains off of species that have a certain number of research-grade observations from different users; species that don’t yet have enough research-grade observations and higher taxa don’t “exist” as far as the computer vision is concerned
There was word back in the old version of the forum that fixes were underway for these problems. It’d be nice to have a sense when these changes might come out if at all.