Ferris Jabr tests iNaturalist Computer Vision on the new Animal Crossing game

This whole thread is fun.


Ferris also wrote this nice piece about iNat for the New York Times back in 2017: https://www.nytimes.com/2017/12/06/magazine/letter-of-recommendation-inaturalist.html


Didn’t mean to delete that prior comment – sorry!

Very fun. Of course, it doesn’t recognize those semi-rare tropical species due to lack of iNat observations in those cases.

Tiger butterfly is a good match for anise swallowtail, which is nearly identical to the model species in the game. The actual species in the game is modelled after Papilio machaon, rather than a “tiger swallowtail” which is a US group of species. This insect in these games was established in the original Animal Crossing game “Animal Forest” which featured Japanese-centric bugs, and when the game was translated to English the meaning was a little misread for localization purposes. Ironically, the species of inspiration in Japan also occurs in North America.

A similar case is the reason for the morpho being labelled as “emperor butterfly” in Animal Crossing, despite this name never being used in English for that species.

This game series was never expected to be popular outside of Asia, and so it’s lead to some curious cultural matches like this when it finally did come to the west.


Yeah, that’s a problem we have where I work in SE Asia. Even with common species at times.

The other problem is Agrias butterflies are usually photographed with the wings shut, so photos showing the upper patterning are not recognized in the database very readily.

“Training iNaturalist’s Computer Vision using screenshots from Animal Crossing: A step by step guide”

Along similar lines I tried using it to identify a soft toy hummingbird:



This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.