This moth is in the vision model but inaturalist doesn’t offer it as a possibility. I don’t think I’ve ever seen the app not offer something that’s correct along with other options. Is it because there aren’t enough adult s observations.
Here is an observation… https://www.inaturalist.org/observations/380718875
When I look at that observation, I get “We’re pretty sure this is in the genus Catocala”, then of the 7 options below that, 6 are species within that genus.
But it doesn’t even list the correct id. But it lists the look a like catocala abbreviatella. I should mention I’m using the android app.
I could understand it not being listed if it wasn’t in the vision model but it is.
Jeff Hansen
https://www.youtube.com/BackyardBirdsUS
I think the linked observation might actually be abbreviatella (the reniform spot is light brown with a dark edge, while nuptialis has a solid dark reniform). That being said, your previous two nuptialis obs look to be correct, with the solid dark reniform:
https://www.inaturalist.org/observations/380642283
https://www.inaturalist.org/observations/380296317
I agree it’s odd that nuptialis never comes up as a suggestion though. It’s in the CV model, and you’re within the “expected nearby” area of the GeoModel for it. But I’ve experimented a bit and can’t get the CV to suggest “nuptialis” for anything. I tried test observations with a few photos of my nuptialis specimens, then I went through and clicked to get a CV suggestion on all the current obs confirmed as nuptialis, and CV won’t suggest nuptialis for any of them- it always picks a different one of the Amorpha-feeders. I tried on the classic app, new app, and browser. Must have something to do with the selection of only ~50 obs being split about 50-50 between larvae and adults.
Sorry, I misinterpreted your query as being about the genus rather than the exact species.
That’s okay. I find it odd that two very similar species aren’t offered
Jeff Hansen
https://www.youtube.com/BackyardBirdsUS
Thanks Paul. And thank you for testing this. I’ve done thousands of observations of things and usually I get the correct I species offered even if it’s down the list but not on this. I have other things with fewer observations that seems to work. Hopefully an inaturalist expert will give us an answer.
Jeff Hansen
https://www.youtube.com/BackyardBirdsUS
CV is 81% sure it is that sp. If you click Visually Similar I get that sp.
Left at the default for obs or RG that sp is bottom of the list – due to less obs.
I never find the default option useful. There are 500 daisies observed here, and these are the 10 most common ones. Gee, thanks, not.
Did you get Catocala nuptiallis. I never see that. I always see catocala abbreviatella. They look very similar. Both are in the vision model. Both should be displayed.
Jeff Hansen
https://www.youtube.com/BackyardBirdsUS
nuptialis has only 55 obs – but is already Included in CV
abbreviatella has 87 obs and is also Included.
But if you are late to the party, and add an ID at the bottom, the right sp is now offered.
For your county (that is also a filter you can toggle for Compare)
https://www.inaturalist.org/observations?place_id=498&preferred_place_id=113055&taxon_id=83842&view=species
It is 5 nup against 2 abb – so that is the cause
What is the cause? I’m completely confused. Nuptialis and abbreviatella should both be offered as they are visually similar. But I never see nuptialis
Not only are ~half the observations larvae, but it seems the larval ones tend to have many more photos included than the adults on average, so if you go into the photo browser the overwhelming majority of photos are larvae. My guess is that the CV has only learned to identify the larvae (and indeed, I tried a few random larval observations and after clicking “include suggestions not expected nearby” nuptialis appeared in each of their suggestions). It is quite common for the CV to be hopeless on lepidopteran larvae and fine on the adults, so this seems to just be an inversion of that due to a skewed dataset.
I really wish the model would separately train and be tested on adults and larvae when possible, treating these as though they were two separate taxa.
I think an iNat expert already gave you an answer!
I’ve added many adult catocala nuptialis but I still have to manually identify them. There are 30 adult observations of the 62 total observations. I’m thinking the çv would be learning this by now.
Jeff Hansen
https://www.youtube.com/BackyardBirdsUS