Visualization of how artificial intelligence has learned over 350,000 flower photos

Infants learn the names of things when adults teach them the names of the objects they see. The latest artificial intelligence research makes it possible to teach computers the name of things by showing many examples, just like humans do. The keys are a large amount of training data and deep learning. By leveraging this latest technology, we have developed artificial intelligence capable of classifying flowers by using over 350,000 examples of flower pictures. This system can now classify 770 kinds of flowers.
When it photographs a flower from a book, it identifies the name of the flower and shows its class among similar flowers on a visual “flower map.” These books contain images of flowers with different abstraction levels, such as pictures, paintings, etc. Observe how it reacts with different abstraction levels.
The web version of “hananona” is available from your smartphone.
Take a picture of flower objects in the surrounding display cases and try the “hananona” web version.

Photo by Surface&Architecture
Photo by Surface&Architecture
Photo by Surface&Architecture
Photo by Surface&Architecture

Stair Lab. collaborated with
Surface & Architecture Inc.(creative direction、design)
Kyoko Kunoh(art direction)
Tomohiro Akagawa(interaction design、programming)
Tanoshim Inc.(programming)
mokha Inc.(server programming)
Tokyo Studio Co.,Ltd.(furniture and fixtures)