Visual data mining of Google identifies cities’ distinctive details.
What makes Paris, the city of lights and romance, so special? Geographically, there is nothing inspiring about it: the city is simply a flat piece of land with a river running through it. What is so impressive about the renowned capital are the manmade creations that go beyond landmarks, such as the Eiffel Tower or Notre Dame, which give the city its allure.
Researchers at Carnegie Mellon University (CMU) and INRIA/Ecole Normale Supérieure in Paris have developed visual data mining software that can automatically detect these subtle features, such as the shape and colour of street lamps and balcony railings, that give Paris and other cities a unique image.
The researchers randomly selected over 250,000 visual elements from Google street view and analysed them. They covered some of the world’s most visited capital cities, including Paris, London, Barcelona and New York. The images were used to determine which details made them different from similar visual elements from other metropolis. They identified that for Paris, the top-scoring visual elements corresponded to doors, balconies, railed windows, street signs and Parisian lampposts.
“We let the data speak for itself,” says Abhinav Gupta, assistant research professor of robotics at CMU’s Machine Learning Department.
The researchers conclude that the look and feel of a city rests not so much on the few famous landmarks, such as the Eiffel Tower or, in the case of Sydney, the Opera House, but largely on a set of stylistic elements, the visual minutiae of daily urban life.
Source: ACM Transactions on Graphics
