Projects

Exploring how AI, geospatial data, and emerging digital data can help us understand cities, populations, and places.

AI for Urban Land-Use Mapping

Nairobi, Kenya

Revealing Digitally Invisible Groups

Urban Planning · Machine Learning · Multimodal Data

Combining satellite, building, social-media and street-view data to examine how different urban environments are represented in AI-based land-use classification.

Revealing digitally invisible groups with multi-source machine learning

Predicting Internet Connectivity

Kenya & Mexico

Can We Predict Internet Connectivity?

With the International Telecommunication Union

Urban Planning · Machine Learning · Spatial Analysis

Testing how well machine-learning and spatial models predict internet connectivity within cities and across countries.

Estimating Population from Satellite Imagery

Gaza

Satellite-Based Tent Detection for Population Estimation

Real Good Research · Ongoing

Population · Computer Vision · Remote Sensing

Using computer vision on satellite imagery to detect tents and estimate population where conventional data are scarce.

Planet Labs satellite image of Gaza Detected tent points overlaid on satellite imagery
Move the slider to show or hide detected tent points.

Measuring How Policy Changes Migration

Chile–United States & Turkey–Germany

Detecting Migration Surges with Digital Trace Data

Mobility · Machine Learning · Digital Trace Data

Using digital trace data to measure how policy changes shift cross-border migration, including onward flows about 240% higher for Chile→US and 122% higher for Turkey→Germany than synthetic controls.

Detecting Land-Cover Change

Peak District, United Kingdom

Land Cover Change Detection

Funded by The Alan Turing Institute

Environment · Remote Sensing · Spatial Analysis

Using deep learning and satellite imagery to detect land-cover change.

Land Cover Change Detection

National Conversation

United Kingdom

Funded by the BBC · Ongoing

Urban Planning · Spatial Analysis · NLP

Linking the places people like and dislike to neighbourhood deprivation, local amenities, and residents’ comments.

Hard at work generating the results.