Using deep learning to identify (urban) form and function in satellite imagery
The case of Great Britain

Dani Arribas-Bel
Martin Fleischmann
[@darribas]
[@martinfleis]

“The point”

How we arrange “stuff” in cities matters…

Source: A map of every building in America (New York Times)

… it matters a lot

Form & Function

Form

What does it look like?
“Physical structure and appearance of cities”

Function

What is it used for?
“Activities that take place within an environment”

Spatial Signatures

A characterisation of space based on form and function designed to understand urban environments

British Signatures

BRITISH SIGNATURES

Countryside (3)

Periphery (4)

Urban (9)

urbangrammarai.xyz/great-britain/

94% (50%)

5% (40%)

1% (10%)

 🛰

Sentinel 2

What do we want to do?

train a neural network
understand the role of geography

Exploration

Chip size effect

80x80m: 13760 chips, which is 74 % of maximum

160x160m: 2718 chips within, which is 57 % of maximum

320x320m: 423 chips within, which is 35 % of maximum

640x640m: 38 chips within, which is 13 % of maximum

320x320m, chips capturing the proportion (100% of maximum)

Current work

Image classification - Overall accuracy 42.8%

Multi-output regression - Overall accuracy 43.5%

Image classification - Wild countryside

Multi-output regression - Wild countryside

Image classification - Urbanity

Multi-output regression - Urbanity

Probability modelling

*the accuracy is based on a different sample than in previous cases (WIP)

Feedback?

  • Better (spatial) evaluation of model performance
  • Probability modeling: does it make (any) sense?
  • Anything else?

Using deep learning to identify (urban) form and function in satellite imagery
The case of Great Britain


Martin Fleischmann
Dani Arribas-Bel
[@martinfleis]
[@darribas]
m.fleischmann@liverpool.ac.uk
d.arribas-bel@liverpool.ac.uk

Great Britain

Characters

Form
  • Dimension
  • Shape
  • Intensity
  • Connectivity
  • Diversity
Function
  • Population
  • Employment
  • Industry
  • Land use/cover
  • Amenity access
N ≈ 300

Context

Data

Form
  • OS OpenMap
  • OS OpenRoads
Function
  • (Business) Census
  • OpenStreetMap
  • Geolytix
  • Listed buildings
  • CDRC
  • CORINE / Sentinel 2
  • VIIRS

Distribution/co-occurrence

Urban hierarchy