Flagger | Content Author | Content | Reason | Flag Created | Resolved by | Resolution |
---|---|---|---|---|---|---|
bouteloua | New Zealand sow thistle (Sonchus kirkii) |
overconfident computer vision suggestion |
May. 12, 2020 23:25:42 +0000 | bouteloua |
see ID link |
Welcome to the world where few species are on AI in a region, but there are good records elsewhere in the world.
What else can one expect from naive users?
I am wondering if the AI (computer vision) should only be done for species with an Atlas. Then the computer vision can flag identifications out of known range -warning users to double check the ID.
[[[[and perhaps offer the viewer the option of marking it as planted/captive (which iNaturalist should not do automatically without at least a warning, as it sometimes does), or create a flag on the atlas that it needs to be revised. ]]]
But I have no idea what proportion of species actually have an atlas ...
Currently 117 observations outside of its known distribution:
https://www.inaturalist.org/observations/identify?not_in_place=6803&quality_grade=research,needs_id&taxon_id=406087