Data and code
Open resources from our group
We believe in open science and reproducible research. Besides open access publications, we are committed to publishing our code and data on our Github account.
Please see below for a list of resources and projects.
If you use the code and/or the data for presentations and publications, we kindly ask you to cite the related paper(s) and credit our work. We have provided guidelines to do so in each resource.
While a great deal of effort has been put into each project, they are not free of errors, and we cannot be held responsible for the use of data and any issues that may arise.
Feel free to contact us for more information, report bugs and errors, or simply to inform us what are you using the data for. We would be pleased to hear how others are using our research.
In you are interested in collaborating with us, please check the opportunities tab for more information.
3D dataset of all public housing (HDB) in Singapore
| Short description: | About 12k semantically rich 3D buildings in Singapore in CityJSON and OBJ |
| Formats | CityJSON, OBJ |
| Further reading: | Please read more here in our blog post or in the paper |
| Download: | Github repo |
| Code: | The code used to generate the dataset is available in a separate Github repo |
| Main data source(s): | HDB, OpenStreetMap, OneMap. All open data |
| Coverage: | Singapore |
Roofpedia – solar and green roofs around the world
| Short description: | Locations of buildings that have installed photovoltaics or greenery on their rooftops |
| Formats | GeoJSON |
| Further reading: | Please read more here at our website or in the paper |
| Download: | Github repo |
| Code: | The code used to generate the dataset is available in the same Github repo |
| Main data source(s): | Various satellite imagery, OpenStreetMap. All open data |
| Coverage: | 17 cities around the world |
Classification of Urban Morphology with Deep Learning
| Short description: | Software to generate diagrams of the urban form at the city-scale and classify them using deep learning |
| Further reading: | Please read more in the paper |
| Code: | Github repo |