I’m wondering out of all the taxa on iNaturalist which are the largest but also most complete? This means which genus, families, orders etc might have the most percentage of their species included in the AI’s training model.
Are there any decently sized families or genus that have every species included in the iNaturalist AI’s training model? If not are there any that are close to being fully incorporated?
I’m also wondering which taxa are the least complete?
Well, any monotypic taxa that’s recorded on iNat is obviously going to top the charts in completeness.
Given the utterly absurd number of species of wasps and beetles, let alone things like nematodes and an enormous range of bacteria, I’d vote for those as the lowest.
Or extremophile deep crust chemoautotrophs. I don’t imagine that many of those are recorded yet, and that’s volume of space vastly larger than the entire surface of the earth with a largely unknown number of species.
I imagine any deep sea taxon will be very incomplete. Not only is species diversity higher than in the above layers, it’s also very expensive to go and record them. So even if iNat has a bunch of deep-sea biologists uploading their findings (or billionaires with expensive hobbies), the user base with access to the abyssal zone or the deep sea benthos remains extremely small.
Even meiofauna isn’t really observed too much and the requirement for that is “just” a microscope (as opposed to an entire research vessel with crew, ROVs, and everything).
That being said, occasionally organisms do get washed ashore so it would be interesting to have an iNat-project with deep sea-creatures. There doesn’t seem to be one for all taxa yet.
I modified the title a bit to make it clearer you were speaking specifically of the computer vision model. Feel free to adjust it
Sure enough, in the plant world class Ginkgoopsida is one of the most complete. 
Monotypic genera, such as Sanguinaria, the bloodroots?
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.