I don’t feel like I can predict what the CV will suggest. Here is an example. I narrowed this one down to a subgenus of Camponotus. The CV makes two suggestions, but look at the members of the subgenus in the region. Why is lasiselene left out? Only two of the ants that have been observed match well at all. Why not just show all of the members of the genus observed in the area?
Of course, the subgenus really contains many more options:
cubangensis
fletcheri
husseini
lasiselene
liogaster
mayri
mendax
mutilarius
opaciventris
pseudolendus
puniceps
sankisianus
scabrinodis
selene
sericeus
siamensis
wasmanni
yiningensis
And many of these have never been observed, so to do a real ID I’d need to look at all of them (some could be skipped if not in the region.) With this ID it’s not that important. The photo isn’t clear enough to narrow to species–
But, I still find the species picked by the CV a little odd. Sometimes the CV saves a little time, but I wish it were more predictable. For example if fewer than 10 species have been observed nearby show them all.
Or always show close visual matches nearby.
Ive noticed the same thing! I often have to flip between ‘observations’ and ‘checklist’, have to turn up the area, etc. Im curious what qualifies something to show up in ‘observations’!
Camponotus lasiselene currently has 15 observations in iNat, so it is impossible for the CV to suggest it, as it isn’t in the model yet. See https://help.inaturalist.org/en/support/solutions/articles/151000170368-which-taxa-are-included-in-the-computer-vision-suggestions-
Many (most? I didn’t check them all) of the other species you also listed above are in the same boat: they currently have few observations in iNat, far fewer than the threshold required for them to enter the model, so it is literally not possible for them to be suggested by the CV (and indeed some of the species you listed above have zero observations on iNat)
As I understand it, the “compare” feature also draws on different data than the CV proper. Depending on what settings you have chosen, this may be observations in the selected taxon seen in the area rather than visually similar matches.
Yeah, Thomas answered the question but FWIW your screenshot shows Source: Observations instead of Visually Similar (CV).
Which yields those two species: https://www.inaturalist.org/observations?place_id=27837&subview=map&taxon_id=795105&view=species
