Platform: INaturalist
App version number, if a mobile app issue:
Browser, if a website issue:
URLs (aka web addresses) of any relevant observations or pages:
Screenshots of what you are seeing:
Description of problem:
Whenever there’s a grape leaffolder or a grape leafroller moth, the AI never suggests the complex. Since most people don’t know about the complex, they just pick the suggested one. There’s tons of posts without enough evidence promoted to research grade. The AI doesn’t seem to be able to suggest the complex at all, as it happens on every post and I believe every device. I have changed around 200 posts marked as grape leafroller or leaffolder moths and changed the ID into the complex by now. Most of that was done in only 4 hours total. There are way more than 200 that have the wrong ID. I don’t know if this belongs here, but it it an issue. It doesn’t recognize that the top is identical. It should be suggesting the complex along with the others.
This is correct, and it’s not just limited to this species complex. The CV is trained only on “leaf taxa”, meaning if a taxon (like a complex) contains another taxon (like a species) which the CV knows, it is no longer trained to recognize the parent taxa (the complex). Enough funeralis/maculalis have been correctly identified based on ventral views or size measurements that the the species are in the CV training, so the complex cannot be suggested any more. There have been quite a few conversations on the forum recently about the possible benefits of training the CV on certain “parent taxa” if there are enough RG observations that are stuck at that higher level.
Also side-note if anyone reading this is cleaning up Desmia, be on the lookout for D. subdivisalis, which often gets misidentified as maculalis/funeralis complex. Males lack the white band at the base of the abdomen that male maculalis/funeralis have, and also tend to have all-white fringes, while the other species have some black patches in the fringe:
https://www.inaturalist.org/observations/390612178
Female subdivisalis have the same abdominal pattern as maculalis but differ in the all-white fringe:
https://www.inaturalist.org/observations/249953424
Supposedly, D. subdivisalis also has more of a purplish sheen, but I haven’t personally been able to notice the difference in sheen between photos of subdivisalis and photos of funeralis/maculalis. And some D. subdivisalis show a darker halo around the white spots on the wings. Slightly more noticeable on D. subdivisalis than funeralis/maculalis.
