Hello iNat community,
Would anyone happen to know if there is a link to the complete list of species that are included in the latest CV model?
I know that you can find out individiually for each species by looking at their “About” page but I was thinking more on the likes of a complete list, preferably in a tabular format.
Thank you !
A.
there are over 1 million species set up in iNaturalist and over 100000 taxa are included in the latest main model. it’s unlikely a human could read through such a large list. so why would you want such a list, particularly in a tabular format?
there’s some related discussion at: https://forum.inaturalist.org/t/searching-for-taxa-that-need-observations-to-be-included-in-computer-vision/35271.
Hello and thank you for your reply.
My work involves giant datasets that I process automatically so actually 100K is not a big deal.
Another approach would be to find out whether the species from a list I am using are included in the computer vision. However, I thought, if the CV list is easily accessible somewhere, it would be faster to download it as I could then subset it myself.
But thank you for your link, seems like I can get some answer from there, I am just not very advanced in that sort of coding.
The blog post contains (examples of) species that are added to the Cv, but a complete list….
There is no ‘complete’ list. But there are links in each blog post for the latest CV update. Since that in turn is difficult to find, I bookmark them as they roll in.
CV 2.23 June 2025 trained off data exported on June 22, 2025. With the explanation of that recent change. https://forum.inaturalist.org/t/cv-suggestions-have-gotten-much-much-worse-at-san-jacinto-mountain/68136/82
We should be close to the next update, since that was published 16 August.
that’s how i would approach it.
nowadays, the v2 API also allows you to explicitly specify the fields that are returned in the response. so you could get the results, 30 taxa at a time, with something like GET https://api.inaturalist.org/v2/taxa/62741,54522,78836?fields=id,name,rank,vision
… which returns something like this:
{"total_results":3,"page":1,"per_page":30,"results":[{"id":62741,"name":"Rudbeckia hirta","rank":"species","vision":true},{"id":54522,"name":"Pisum sativum","rank":"species","vision":true},{"id":78836,"name":"Ratibida columnifera","rank":"species","vision":true}]}
There may be some lingering bugs with the API’s vision field – i.e. I’m not 100% confident you can trust the result, see https://forum.inaturalist.org/t/cv-model-included-flag-bug/47991/12
However, you should be able to get a basically complete list from paginating through https://api.inaturalist.org/v1/observations/species_counts?expected_nearby=true&page=1 (also available as a v2 endpoint)
oh, that’s an interesting behavior for the expected_nearby parameter. i guess then it’s possible to export a nearly complete list of taxa using https://jumear.github.io/stirfry/iNatAPIv1_observations_species_counts?expected_nearby=true&options=unlimitedexport.
it looked like the GitHub issue was resolved and closed. it was tied to yours and another bug report in the forum: https://forum.inaturalist.org/t/taxon-page-computer-vision-model-status-affected-by-place-filter/59759. this other one got closed, but yours is still open. i’m not sure if that means an issue still exists though.
Yes, that’s why I said “may be some lingering bugs”
I use this, I’ve mentioned it before a few times in other posts. It is very close to perfect like 99%, *not fully though. It is less accurate right after a CV update, but gets better until the next and it cycles.
It is configurable to location and taxon. Make both blank for a global list. Includes any leaf taxa, not just species as long as its in the CV.
Chironomids (Non-biting Midges) have 79 taxa learned.
Biting Midges have 6
Phantom Midges also have 6
Yes, the “expected_nearby=true” parameter actually seems to be the solution! And the jumear link is perfect for downloading the complete list. Thank you!