A team of scientists from the University of Toronto has developed A method that uses artificial intelligence (AI) and Google Maps Street View images For more detailed information about buildings, such as their age and built area.
These additional data can be used to evaluate the real estate park, The flows of construction materials and the greenhouse gases incorporated, which are estimates of the emissions generated by the production and transport of goods. The results of this analysis have been published in Journal of Industrial Ecology.
“This is the first article we know in which a photo was taken that shows the facade of the building and then information is predicted that cannot be seen in the image -explains Shoshanna Saxe, leader of the study, in a statement -. My motivation focused on the use of the investigation into the incorporated carbon, but this will be useful for many people. I have spoken with researchers who seek to understand the use of water for future planning or resilience evaluations. ”
Thanks to the wide availability of Google Street View, The method offers a profitable way to generate large -scale buildings data.
“We spend about $ 1000 on photos to obtain data that, otherwise, it would cost millions of dollars – adds Saxe – no one has millions of dollars to invest only in the dimensions of the buildings, So this is the difference between being able to work on these problems or not. Having methods that allow us to understand neighborhoods and buildings on scale is really useful. ”
The team trained the AI to estimate the building’s attributes based on exterior images of the structure, achieving a accuracy of 70 % in the prediction of age and 80 % in the prediction of the constructed area.
“Evaluating the exteriors allows a founded estimate of the interiors and the type of use that the occupants give to the local infrastructure -explains co -author Alex Olson -. Offers a solid estimate of the resources used in constructionthe maintenance and operation of the buildings ”.
Saxe adds that The information obtained with your approach cannot be obtained solely from maps or construction plans. It is necessary to see the structures according to the authors and one of the differences is that we predict the inner surface of the building. And, although obviously this coincides with the external size, it is actually more difficult to predict. In addition, you cannot see the age of the building from the outside.
“If you have experience, you can travel it And to say: “That building seems more or less like that, this other seems more or less like that, and so on.” But there are all kinds of factors that make it difficult, including renovations. The facade can be different from the rear. And the facade is brick, glass or concrete? Knowing the age of the building is important, since it indicates what materials they were used and what incorporated carbon contains. And also his performance, ”says Olson.
It is true that using AI to see beyond the facades of buildings could help urban planning experts to better understand cities resource needs and prioritize future infrastructure projects. But also It provides important information for engineers when making controlled reforms or demolitions and, obviously, in military objectives.
“It is important to understand where there are infrastructure resources or infrastructure in the city -concludes Olson -. It seems that we should already have the data, but we don’t really have them. With this, with this, While the future is not modeled, the current situation is described quite accurately and we are allowed to use the data to plan the use of resources and what we want to do in the future ”.