Artificial intelligence and misidentifications

[bouteloua]

@djpmapfer

I want a system where I as an informed user am presented with a proposed identification and I can respond to say that the proposal is incorrect (indeed massively incorrect).

See the discussion on this related feature request:

@bouteloua

I checked with the staff on this one and it doesn’t look like they’re going to move forward with adding a different kind of flag or alert system for computer vision training errors. I’d recommend adding species here: https://forum.inaturalist.org/t/computer-vision-clean-up-wiki/7281 and using the existing @ mention and flagging system to get help with people reidentifying incorrect observations, which in turn helps retrain future versions of computer vision.

Ken-ichi had a nice post yesterday partly about computer vision and the thresholds for accuracy/number of training photos. Might be a good place to ask some follow-up questions: https://forum.inaturalist.org/t/identification-quality-on-inaturalist/7507