Hello, I was reading a pre-proof that references iNaturalist and they refer to old papers on the iNat 2017 computer vision challenge that accounted for ~8000 species being identifiable based on fine-grained image analysis. I’m under the impression that the model has grown a lot in its capacity to recognize more species since then, and am unsure of how many species can be identified using the model at the present time.
This isn’t a point of correcting or changing the pre-proof, just a means of getting an updated statistic since we now have over 25M observations and 200,000 species on the platform.
Might be a good idea for a blog post, honestly.