/* Additional CSS for hover effect */ .hover-title { position: relative; } .typewriter-text { /* Hide the text initially */ display: inline; visibility: hidden; opacity: 0; color: white; /* Set text color to white */ transition: opacity 0.3s ease-in-out; } .typewriter-link:hover + .typewriter-text { /* Show the text on hover */ visibility: visible; opacity: 1; }
︎
Carmelo Ignaccolo
Team

Nick Sundholm 
Shuman Wu

Datamining the City



This machine learning project wants to respond to the need of city mass models when there is no other data available apart from Google Street View.

The CNN uses as dataset 32,000 images from Manhattan collected with an algorithm from Google Street View. The data has been matched with the GIS data available in NYC for the height of the buildings in order to train the network and test its accuracy. The approach towards the CNN is a supervised classification.

The visualization is the result of a combination of tools such as Rhino, Grasshopper, Meerkat and Arcmap.