Urban Computing to Gain Insight into City Liveability

The urban computing to gain insight into city liveability solution focuses on developing cutting-edge AI techniques that allow us to gain insight and extract information from multiple data sources, all in the context of analyzing urban liveability.

The challenge

More than half of the world’s population lives in urban areas, and it is expected to increase to two-thirds by 2050. Cities are faced with many big challenges, ranging from fighting air pollution to providing affordable housing for all, and they are looking for new, innovative ways to tackle them. Due to the complex and dynamic nature of our cities, knowledge from different research fields needs to be combined. In this context, urban computing shows great promise in providing potential solutions. This interdisciplinary field, connecting computer science to more traditional city-related areas, such as environmental sciences, sociology, and economics, focuses on acquiring, integrating and analyzing big data in urban spaces. In this project, we use urban computing to analyze different aspects of urban liveability. The aim is to develop a method that allows to objectively measure and evaluate the main contributing factors to city liveability in different parts of the city. The project aims to answer:

  • How come a specific area within the city is perceived as either safe or unsafe?
  • How does the design of public space influence citizen health?
  • What are the key aspects that influence the demographics of an urban area (and vice versa)?

The solution

This solution focuses on developing cutting-edge AI techniques that allow us to gain insight and extract information from multiple data sources, all in the context of analyzing urban liveability. Data sources are not restricted to typical statistical data, but also consider visual data sources such as panoramic street view images, textual data streams such as Twitter messages or service requests, as well as information from knowledge bases of the municipality.

The team

This solution is led by the Amsterdam Data Science (ADS) team, which involves Inske Groenen, Dr. Stevan Rudinac, and Prof. Marcel Worring of the University of Amsterdam. At present, the municipality of Amsterdam is the main data provider for this solution. At a later stage in the process, other cities will be able to step in to test and implement the solution in their city. Multiple cities have already expressed their interest in doing so.

The current status

In previous work, the work was focused on generating discriminative visual summaries of geographic areas by utilizing user-contributed images, and on identifying geographic regions within the city based on social multimedia and open data. We also analyzed what multimedia sentiment can tell us about city liveability. In addition, we have been working on interactive multimodal recommendation systems for guiding tourists outside of main tourist zones. Finally, in our recent work we have developed a multimodal approach to detect urban microevents, which has been deployed for automatic classification of citizen reports in the City of Amsterdam. Now, we focus on identifying and localizing common objects within panoramic street view images. To this end, we have developed an approach to get an approximation of which objects are visible in which image, and where, based on information from Geographic Information Systems (GIS) databases. These approximate image labels will enable the algorithm to learn the task of identifying and localizing urban objects. In order to accurately evaluate the performance of our algorithm, we are now working on getting a subset of the images annotated with absolute accurate information on which objects are visible in the image and where. Annotations will be collected via crowd sourcing, and we have just released an online tool to start collecting these human annotations. The dataset that is being developed will be released as a new open urban dataset with benchmark tasks relevant to the computer vision and multimedia analytics community.

Urban liveability encompasses many of the challenges for which SCORE aims to provide smart, open data driven solutions. Insight into urban liveability is the first step in effectively tackling the open challenges, and AI techniques can be a very effective and efficient way to gain insight. In addition, we are releasing an open dataset to advance both computer vision, multimedia analytics research and public service delivery.

Ambitions after SCORE

We aim to provide a dataset and develop algorithms that will inspire further research even after the end of the project. In addition, the tools developed within the scope of this project aim to help public servants in the future.


Inske Groenen - i.groenen@uva.nl

Dr. Stevan Rudinac -  s.rudinac@uva.nl

Prof. Dr. Marcel Worring - m.worring@uva.nl