How AI Sees the City: Urban Visual Intelligence and the Future of Urban Planning
- Artificial intelligence is emerging as a critical tool for observing and analyzing urban environments, offering new ways to measure vehicle emissions, track public space utilization, and evaluate green...
- Researchers at the Massachusetts Institute of Technology Senseable City Lab released a book titled 'AI가 도시를 보는 방식: 도시 시각 지능(How AI Sees the City: Urban Visual Intelligence)'...
- A primary example of visual AI application highlighted in the research involves air pollution studies in New York City.
Artificial intelligence is emerging as a critical tool for observing and analyzing urban environments, offering new ways to measure vehicle emissions, track public space utilization, and evaluate green infrastructure. By processing visual data generated across city streets—including footage from traffic cameras—researchers can extract precise metrics that inform modern urban planning and public health policy.
MIT Researchers Publish Guide on Urban Visual Intelligence
Researchers at the Massachusetts Institute of Technology Senseable City Lab released a book titled ‘AI가 도시를 보는 방식: 도시 시각 지능(How AI Sees the City: Urban Visual Intelligence)’ on Sept. 24, exploring the applications, possibilities, and risks of visual AI in urban research and design. Published by Routledge, the volume examines how computer vision translates physical city elements into quantifiable datasets. The book is authored by Fábio Duarte, Martina Mazzarello, Carlo Ratti, and Fan Zhang. The authors explain that visual AI converts diverse urban features into digital data to analyze operational patterns and resident experiences. However, they also emphasize that widespread video collection and analysis raise serious concerns regarding privacy invasion and algorithmic bias, requiring a cautious approach to technology deployment.
Monitoring Vehicle Emissions Using Traffic Cameras
A primary example of visual AI application highlighted in the research involves air pollution studies in New York City. The MIT Senseable City Lab team utilized machine learning to analyze video feeds from 331 New York City traffic cameras, identifying vehicle types on roadways and estimating individual automobile emissions.
This methodology demonstrates that comprehensive camera coverage can enable cities to monitor vehicular emissions more precisely and extensively than traditional stationary sensors allow. Beyond traffic and environmental analysis, visual AI extends to identifying specific causes of traffic congestion, pinpointing intersection accident risks, and evaluating how people use public plazas and parks.
In urban planning, visual data functions as an analytical asset rather than a static record. Applying computer vision allows researchers to isolate and quantify distinct elements in a single image, including vehicles, buildings, pedestrians, and green spaces.
Digital images are treated as data, and the characteristics of the city can be quantified,
Fábio Duarte explained, noting that computer vision technology utilizes each image as an individual dataset.
Scaling Decades of Traditional Urban Observation
The research team views visual AI not as an entirely unprecedented departure, but as a technological expansion of long-standing visual observation traditions. For centuries, physical records—from ancient Roman marble maps to early photography documenting 19th-century Paris and the crowded tenements of New York’s Lower East Side—have served as vital instruments for comprehending urban morphology.
In 1960, Kevin Lynch published The Image of the City,
analyzing how individuals perceive and remember spatial environments. Sociologist and urban researcher William H. Whyte similarly advanced urban design by observing human behavior in public spaces. The Senseable City Lab scales these observational traditions into the digital era, using massive video datasets to evaluate multiple locations and dimensions simultaneously.
Kevin Lynch worked with just paper and pen,
Carlo Ratti stated. Today, we can use visual AI to scale up the work he did and examine various aspects of the city simultaneously.
