TLDR: The ‘Street Review’ framework combines participatory research with AI-based image analysis to assess streetscape inclusivity. Conducted in Montréal, it involved residents’ interviews and image evaluations, correlating subjective ratings with physical attributes like sidewalks and greenery. The study revealed varied perceptions across demographic groups and demonstrated how diverse user feedback can enhance AI models for urban planning. It offers a systematic method for planners to inform policy and management of public streets, generating citywide heatmaps of perceived inclusivity.
Our cities are constantly changing, with new people and cultures shaping how public spaces are used. This makes it crucial to regularly check if our streets are welcoming and accessible to everyone. Traditional ways of evaluating urban spaces often fall short, either by using a ‘one-size-fits-all’ approach that overlooks diverse needs or by lacking the detailed human perspective needed to truly understand inclusivity.
Introducing Street Review: A New Approach
A recent study introduces a new framework called ‘Street Review,’ which combines the valuable insights of local residents with advanced Artificial Intelligence (AI) analysis to assess how inclusive streetscapes truly are. This innovative method aims to provide urban planners and policymakers with a systematic way to understand and improve public streets.
The research, conducted in Montréal, Canada, involved 28 residents who participated in in-depth interviews and evaluated images of streets. This human feedback was then combined with the analysis of about 45,000 street-view images from Mapillary, a crowdsourced mapping platform. The goal was to see how people’s subjective feelings about a street correlated with its physical features, such as sidewalks, maintenance, green spaces, and seating.
How the Study Was Conducted
The ‘Street Review’ framework uses a multi-step process:
-
Gathering Resident Perspectives: Researchers recruited a diverse group of participants, including senior citizens, women, ethnic minorities, people with disabilities, and LGBTQ2+ individuals. They conducted semi-structured interviews, asking residents about their favorite and least favorite public spaces and what qualities made a street work for them. These conversations revealed that different groups had different priorities; for example, people with mobility impairments focused on ramps and sidewalk conditions, while younger participants valued visual interest and flexible use.
-
Defining Key Criteria: From these interviews, four main criteria for evaluating streets emerged: accessibility, aesthetics (how visually appealing a street is), practicality (how functional it is), and inclusivity (how welcoming it feels to all). These criteria formed the basis for further evaluation.
-
Image Rating by Focus Groups: A smaller group of 12 participants then rated images from 20 selected street locations in Montréal based on these four criteria. They discussed their reasoning, leading to a shared understanding and collective ratings for each image. This ‘co-production’ process helped ensure that the AI model would be trained on human-defined perceptions.
-
AI Model Training: The study used a sophisticated AI model called SegFormer to analyze the street images. This model was trained to identify various elements in the streetscape, like sidewalks, buildings, and vegetation. A custom AI model then used these identified features, along with the human ratings, to predict scores for practicality, aesthetics, accessibility, and inclusivity across a much larger dataset of images. The model was designed to understand how different features contribute to these perceptions for various demographic groups.
What the Study Found
The findings showed interesting patterns:
-
Human Perceptions: Participants generally felt that inclusive streets were also accessible and aesthetically pleasing. However, they noted that practical features like ramps didn’t always make a street look more appealing.
-
AI Model Insights: The AI model found a strong link between accessibility and practicality. It also suggested that visual qualities (aesthetics) played a significant role in its estimation of inclusivity, sometimes even more than physical accessibility. This highlights a slight difference between how humans and the AI model ‘see’ inclusivity.
-
Demographic Differences: The study confirmed that perceptions of inclusivity vary greatly. Elderly males, young females, and individuals with mobility impairments often gave lower inclusivity scores, citing concerns about safety, narrow sidewalks, or lack of ramps. In contrast, younger males and LGBTQIA2+ participants often gave higher scores, appreciating vibrant cultural areas and nightlife.
-
Citywide Patterns: When the AI model was applied to 45,000 images across Montréal, it generated ‘heatmaps’ showing that central, well-maintained districts generally scored higher for inclusivity, while peripheral neighborhoods often had lower scores, reflecting infrastructure gaps.
Implications for Urban Planning
The ‘Street Review’ framework offers valuable tools for urban planners. It emphasizes that simply building sidewalks or bike lanes isn’t enough; understanding the nuanced experiences of diverse groups is key. The study suggests that investments in sidewalk infrastructure and crosswalk design can boost perceived inclusivity, and aesthetic improvements like landscaping can further enhance a sense of welcome.
Unlike other automated tools that might miss the human element, Street Review integrates community feedback directly into the AI’s learning process. This helps to reduce biases and ensures that the assessments are context-specific and reflect the lived experiences of residents. While the AI can efficiently analyze large datasets, the human input provides critical qualitative information about cultural identity, historical context, and emotional resonance that purely quantitative metrics cannot capture.
Also Read:
- Mapping Urban Mobility: How AI and Street View Images Predict Cycling and Motorcycling Trends
- Daily Greenery Exposure Linked to Significant Health Benefits in Urban Areas
Looking Ahead
Despite some challenges, such as the resource-intensive nature of co-production and limitations in image quality from crowdsourced platforms, the ‘Street Review’ framework provides a practical and scalable way to conduct inclusive audits of urban spaces. Future improvements could involve more participants, integrating real-time data like footfall counts, and refining algorithms to better detect subtle cultural symbols.
By combining community input, image ratings, and advanced AI, ‘Street Review’ helps identify features that make streets welcoming or unwelcoming, guiding investments toward creating truly inclusive urban environments for everyone. You can explore more about this research at the research paper.
