Modelling preferences on street redesigns. A perception-driven framework using visual and segmentation data

November 21, 2025

Urban streets are increasingly redesigned to prioritise sustainable mobility, yet planning processes often overlook how citizens visually perceive and emotionally interpret these transformations. This gap can generate resistance to otherwise beneficial interventions. To address this challenge, this dissertation developed a perception-based analytical framework linking measurable visual changes in street environments to subjective evaluations. Using temporal images of real interventions, semantic segmentation and perceptual colour metrics were applied to quantify spatial and chromatic variations such as vegetation gain, surface redistribution and hue balance. These indicators were integrated with survey-derived perceptual ratings through factor analysis and structural equation modelling to identify the latent constructs and to predict preference of redesigned streets. Results showed that both colour and segmentation-based predictors captured distinct yet complementary perceptual mechanisms: colour diversity and brightness enhanced liveability and comfort, while balanced functional composition supported perceptions of safety and order. The framework yielded predictions across demographic and behavioural groups, revealing exploratory moderation effects and offering interpretable guidance for design. Despite limitations related to sample scope, illumination variability and segmentation bias, the approach demonstrated the feasibility of operationalising perception within urban analytics. Overall, the study advanced the integration of visual data and human perception into evidence-based planning, providing a replicable tool to anticipate public reactions and guide participatory, sustainable urban transitions.

Authors in the community:

• 

  M

  Miguel Gouveia Pinto Leite Alves

  ist199814

Supervisors of this institution:

• 

  F

  Filipe Manuel Mercier Vilaça e Moura

  ist14188

• 

  G

  Gabriel Costa Valença

  ist194281

civil-engineering - Civil engineering

por - Portuguese

Instituto Superior Técnico