i think if it were me, here’s what i would do:
- open the upload page
- add an observation without media
- open up the firefox web developer tools > network
- load an image to the observation (drag an image file to the observation “card”)
- click the taxon button to initiate the computer vision suggestions
- in the network window, find the two POST and the one OPTION requests. right click on each and save all as HAR.
- send those to email@example.com. to see if they can make anything out of that. (don’t share the contents of the file with anyone else because it will contain authorization keys and other information that someone could use to access your iNat account, among other things.)
if you see errors, too, in the console, make sure you capture those, and send those, too.
i don’t really know why you would be encountering problems only on that screen. i can only guess that maybe iNaturalist is sending you a cached version of the computer vision results that isn’t relevant to your particular observation. if you have access to a VPN, you could try doing your upload through the VPN, and see if that changes the results that you see. (or if you’re already using a VPN, try it without the VPN, and see if that changes things.)