Exploring the composition of Urban Area of Interests through deep learning

Authors: Meixu Chen*, University of liverpool, Dani Arribas-Bel, University of Liverpool, Alex Singleton, University of Liverpool
Topics: Urban Geography
Keywords: Urban Area of Interests, Image Recognition, CNN, Geographic Data Science
Session Type: Paper
Day: 4/3/2019
Start / End Time: 12:40 PM / 2:20 PM
Room: Harding, Marriott, Mezzanine Level
Presentation File: No File Uploaded

Urban Area of Interests (UAOIs) is a concept that provides functional definitions of a city’s spatial structure. Obtaining insights about the contents of UAOIs is crucial to the tourism planning and targeting advertisement. Several new forms of digital data derived from urban activity through passive or active forms of data collection capture urban form and/or social functional geography, such as social media data. Such data have been used to explore spatial, temporal, semantic, and even graphical information about human activities. In this study, we extract UAOIs from geotagged photo data covering the three-year period in Inner London (A series of centrally located London Boroughs within the Greater London Authority Extent), using a Convolutional Neural Network model to understand features within UAOIs and Non-UAOIs. For further understanding the composition differences between UAOIs and Non-UAOIs, we extracted UAOIs in different time slots and applied the model to see if the contents people get interested in would change with time. The research explains why people get interested in these areas and what the general characteristics of UAOIs are. The results would be informative for stakeholders such as urban construction on the built environment by applying deep learning in urban studies.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login