Did you have a look at the paper? They used a deep learning algorithm. Despite not being intelligent in any human sense, humans call these algorithms AI. ^^
Oh, and BTW:
All datasets were normalized to an uniform size of 256 by 256 pixels without any other type of pre-processing. This is the current state-of-the-art resolution as deep learning models are intensive in computing.
Holy shit, Batman - even with ultrasonic vision I wouldn’t be able to ID at this resolution.
I’ll have to look at the details later, but this is definitely interesting. Also:
[W]e did show that it is not beneficial to do transfer learning from herbarium data to leaf scan pictures and it is even counterproductive to do transfer learning from herbarium data to field images.
Not unexpected, but also interesting. Especially that they even tried on fruits. Leaves, ok. But fruits? Noone would identify a banana after looking at a flattened, dried specimen of the fruit, would they?
I remember the Microsoft research division training an AI to identify plant pics from China.