Different computer vision model on upload and observation pages?

I was curious to see recently that the list of suggestions for the same photo was different in the upload page (as follows):

to that I get once the observation has been uploaded (as follows):

The latter was a better suggestion, which makes me wonder if the more recent version of computer vision is being used on the observation pages, and an older version in the web uploader. Or is there another explanation?

1 Like

I’ve noticed that too, but I figured it was taking my ID into account when giving suggestions the second time. If you withdraw your ID, what does it suggest then?

4 Likes

Interesting! Yes, that seems to explain it. When I deleted my ID, the AI suggestions were as when I was in the uploader. I didn’t realise that the suggestions took account of the observation ID. Thanks!

1 Like

Yeah I wonder if it does that only if the ID you put in first wasn’t selected from the suggestions. Otherwise you could go down a rabbit trail of “suggested” names. If you did pick one a second time though, then it would be suggested, and maybe it wouldn’t give you a third list the third time.

1 Like

In the first screenshot, of the web uploader, had the location been entered when you got the computer vision ID?

1 Like

I just tried this again to check, and yes, the location was already entered (it was taken from the photo metadata, and I could confirm that the “loading metadata” dialogue was no longer showing and that the location was present in the observation). I still get Membracinae as the top suggestion there.

1 Like

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.