With the new iNat Geomodel and its “Expected Nearby” feature being added to Computer Vision, it seems like the suggestions it will offer will generally be of higher accuracy and quality. In my 3,000-some observations, many have been left with only very broad taxon-level identifications. Sometimes due to poor photos, admittedly, but sometimes because the Computer Vision couldn’t produce any suggestions that seemed accurate.
Is it worth revisiting some of those old observations to see what IDs that CV can come up with, now that it has been improved?
You know, I had not thought of that, but it sounds like a promising idea to move along some of our older ‘Needs ID” and high-level ID obs. It would be easy to filter for those factors, too.
Given how frequently the CV model is updated, it might be worth it regardless of the new Geomodel. When I’m working through years-old Needs IDs, I frequently run (or rerun) organisms that I don’t recognize but seem visually distinctive through the CV. Sometimes I get a clear species ID, but even if I don’t, I’m still usually able to give a more specific ID than the observation currently has. (Note that those are often from looking at the shared taxonomy of the top few species, rather than the CV giving a “pretty sure it’s in this genus/tribe/subfamily” suggestion.)
I’d be careful here. There are groups where only one sex is visually identifiable, but the CV doesn’t take that into account. A good example is Agapostemon: female Agapostemon virescens are distinct, males of several species are not, but there are a ton of species ID’d Agapostemon virescens males that were from the CV, or identifiers who didn’t know. Just because the picture on the CV “looks like it” doesn’t mean that there aren’t five others that also look it that are also in the same area. We had to deal with that twice in the last month with Sweat Bees, where people were taking the CV suggestion at high volume.
And there are there are groups of Lepidoptera where only one or two species that are included in the CV have caterpillar photos, so the CV tells people that all vaguely similar caterpillars from that group are probably that species. All the usual CV caveats still apply, but it can still be helpful
Another problem is regional variations, Bombus melanopygus has red on the abdomen in the north and black in the south, but taxon picture can only show one. So looking through the CV suggestions without understanding that won’t really help.
Realistically, this change is only going to help with broad strokes, hopefully will cut down on the intercontinental suggestions, but there are a lot of species in a given area, and many look similar and may not look like their taxon picture.
That’s what I thought @neylon probably meant, but I wanted to double-check before I gave the possibly repetitive reminder that if you notice there’s a common variation missing from the set of taxon photos, you can add a photo showing it. And I have to say, I think most people understand that the first taxon photo that shows up as the icon isn’t the only way the organism can look. But I do think it’s helpful to make sure that the “different” taxon photos aren’t all the way at the end of the set.
Just to clarify, any user can add a new taxon photo to its info page if something is missing? Are they curated somehow? How would someone go about doing that?
Or are you referring to simply uploading an observation that shows that variation?
There are loads of insects and other arthropods where there aren’t any taxon photos that display different stages of life (eggs, larvae, instars, etc.), but rather only the adult stage. I’d be happy to contribute photos in these cases if I’m positive about the ID.
Yes, any user can add these, and this is a type of curation. Other users will edit/fix mistakes as they are noticed (and there are plenty, usually from observations that have since been IDed or pics imported from off iNat of misIDed things there like Flickr).
There are loads of insects and other arthropods where there aren’t any taxon photos that display different stages of life (eggs, larvae, instars, etc.), but rather only the adult stage. I’d be happy to contribute photos in these cases if I’m positive about the ID.
Awesome! Yes, this is one of the most common kinds of missing variation for insects. Sometimes it’s just a matter of finding an existing RG observation and adding photos from that to the taxon page, but there are a lot of species where there are no RG observations of non-adult life stages. Sometimes that’s because the larvae, etc. are very difficult to find (e.g., if they burrow deeply into wood or soil) but other times it’s because they’re almost impossible to ID to species with photos.
