Inteligencia artificial para la asignación automática de categorías constructivas

In the article we will present a newly developed algorithm for the automatic assignment of construction typologies using a sample of images obtained from the regional management of the Cadastre of Castilla y Leon, containing buildings of Salamanca town. We only use the images for classification, without taking into account data such as the year of construction, building size or other metadata, as this is how the cadastral technicians work, which makes the problem more complex. We are getting success rates already higher than 85% with a very small sample of images. We use Artificial Intelligence, specifically a deep learning algorithm called Convolutional Neuronal Networks