The landscape paintings are drawn on the canvas by the painters seen and thought in their brain who lived in the real world of that period. But in modern times, the art experience is merely that just looking at the pictures in art museums.

We succeeded in creating a new art experience by connecting the atmosphere between viewers and the artist.

We taught the database by deep learning and created an AI that can estimate the temperature and humidity of the painted world. We have succeeded in estimating “the painting’s world’s” temperature and humidity scores. Those scores of the viewers’ environment are acquired in real-time and entered into the AI. The AI then refers to public domain data on paintings which has been made available by art galleries all over the world. By placing themselves within the world of the painting, viewers can breathe the same air as the painted scenery and experience art in a new, immersive way.

In this project, we had to prepare an abundant sample data as we used deep learning to estimate the scenery's temperature. Then, we built a dataset made by geotagged and time stamped picture about 6700000 samples to unite public domained picture dataset(Yahoo Flickr Creative Commons 100M) and geometroligical dataset(Global Historical Climatology Network)

First, we made possible to suggest by the air and create new way of discovering art by using the AI that estimates temperature and humidity by pictures.

Next, we produced a more dynamic art experience by connecting a real-time meteorological data.