Pardon me if this is the wrong section of the forum. Since a certain time (perhaps a moth or two), the computer vision model identifies Cotoneasters of the bullatus group as Cotoneaster coriaceus (the problem started very suddenly), which is strange since they look nothing alike.
The computer vision model was updated in July of this year.
There really isn’t anything that can be done immediately about those kinds of CV-based misidentifications other than to be extra vigilant in offering corrections and going through old IDs to ensure that they are correct. If lots of incorrect IDs are in the system, then the model’s output will perform poorly.
The added problem for iNat is that the CV model forms a feedback loop, since many users rely on CV suggestions. And again the only way to fix that for a particular group is to ensure that good data gets fed into the system.
2 Likes
Can you provide the URL of an example where this happens? Specific details are really helpful.
2 Likes
- which species are included in the bullatus group? (i don’t see anything in the iNaturalist taxonomy itself that defines such a group.)
- C. bullatus (species) itself is not included in the iNat computer vision model because it doesn’t have enough observations for inclusion. if the computer vision wasn’t recommending, say, C. bullatus (species) as a suggestion, and the iNat taxonomy doesn’t have that group defined, then what would the computer vision have suggested in the past, if not a common species like Cotoneaster coriaceus?
lately, i’ve been wondering whether, as more species get added to the computer vision model, accuracy could get worse over time? if the vision used to know only about 2 species that looked alike and now it knows about 15, and it can recommend only up to 10, then it would seem harder to make a proper recommendation from 15 than 2 possibilities.
May offer us more of
We’re not sure of the species
but it is this Genus
which is better than saying we are pretty sure it is this species 'cos that is the only one we know.
1 Like