Quantifying the Characteristics of the Local Urban Environment through Geotagged Flickr Photographs and Image Recognition
Meixu Chen; Dani Arribas-Bel; Alex Singleton (2020). ISPRS International Journal of Geo-Information, 9(4), 264. DOI: 10.3390/ijgi9040264
Abstract
Urban environments play a crucial role in the design, planning, and management of cities. Recently, as the urban population expands, the ways in which humans interact with their surroundings has evolved, presenting a dynamic distribution in space and time locally and frequently. Therefore, how to better understand the local urban environment and differentiate varying preferences for urban areas has been a big challenge for policymakers. This study leverages geotagged Flickr photographs to quantify characteristics of varying urban areas and exploit the dynamics of areas where more people assemble. An advanced image recognition model is used to extract features from large numbers of images in Inner London within the period 2013–2015. After the integration of characteristics, a series of visualisation techniques are utilised to explore the characteristic differences and their dynamics. We find that urban areas with higher population densities cover more iconic landmarks and leisure zones, while others are more related to daily life scenes. The dynamic results demonstrate that season determines human preferences for travel modes and activity modes. Our study expands the previous literature on the integration of image recognition method and urban perception analytics and provides new insights for stakeholders, who can use these findings as vital evidence for decision making.
Extended Summary
This research investigates how geotagged social media photographs can reveal patterns of human activity and preferences in urban environments. Using more than one million Flickr photographs taken in Inner London between 2013 and 2015, the study applies advanced image recognition technology to understand what makes certain urban areas more attractive to people than others. The methodology combines several cutting-edge techniques. First, the research identifies Urban Areas of Interest (UAOIs) using spatial clustering algorithms that detect where large numbers of people congregate to take photographs. These areas are then compared with other urban locations to understand their distinctive characteristics. The analysis employs Places365-CNN, an advanced deep learning model trained to recognise different types of scenes and environments in photographs, rather than relying on user-generated tags which can be inconsistent or missing. The findings reveal clear differences between popular urban areas and everyday locations. UAOIs predominantly contain iconic landmarks, tourist attractions, and leisure facilities such as towers, bridges, skyscrapers, churches, museums, and shopping centres. In contrast, less popular areas feature more mundane daily-life scenes including bus stations, residential streets, and indoor venues like conference centres and shops. These results demonstrate that people are drawn to photograph distinctive architectural features and culturally significant locations rather than routine urban infrastructure. The research also uncovers important seasonal patterns in urban behaviour. During winter months, photographs in popular areas more frequently show transport infrastructure and indoor activities, suggesting people prefer vehicles over walking and indoor venues during cold weather. Conversely, warmer months see increased photography of outdoor spaces, crosswalks, and recreational areas, indicating greater pedestrian activity and outdoor leisure engagement. Specific events also drive temporary popularity, such as Hyde Park’s Winter Wonderland creating amusement park characteristics in December. This work demonstrates how social media data and computer vision can provide valuable insights for urban planning and policy. The methodology offers policymakers a new tool for understanding how people actually use and perceive urban spaces, moving beyond traditional survey methods to real-time, large-scale analysis of human behaviour. The seasonal patterns could inform transport planning, helping authorities anticipate changing travel demands throughout the year. Similarly, retailers and local councils could use these insights to optimise services, extend opening hours during peak periods, or target advertising more effectively. The research contributes to the growing field of urban analytics by showing how image recognition technology can quantify subjective aspects of city life, offering a data-driven approach to understanding the complex relationship between urban design and human behaviour.
Key Findings
- Urban Areas of Interest predominantly contain iconic landmarks, tourist attractions, and leisure facilities rather than everyday infrastructure and daily-life scenes.
- Seasonal patterns significantly influence human urban behaviour, with winter favouring transport and indoor activities, summer promoting outdoor pedestrian engagement.
- Advanced image recognition models can successfully quantify urban characteristics from social media photographs, providing alternative to traditional survey methods.
- Popular urban areas show distinct visual characteristics including towers, bridges, museums, and shopping centres that differentiate them from routine locations.
- Geotagged social media data combined with machine learning offers policymakers new tools for evidence-based urban planning and transport management decisions.
Citation
@article{chen2020quantifying,
author = {Meixu Chen; Dani Arribas-Bel; Alex Singleton},
title = {Quantifying the Characteristics of the Local Urban Environment through Geotagged Flickr Photographs and Image Recognition},
journal = {ISPRS International Journal of Geo-Information},
year = {2020},
volume = {9(4)},
pages = {264},
doi = {10.3390/ijgi9040264}
}