Reproduce “UrbanGazeVis: A Visualization System for Analyzing Eye-Tracking Data on Urban Safety Perception” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Andres De La Puente, Luis Sante, Felipe Moreno-Vera, Mauro Diaz, Jorge Poco
UrbanGazeVis links street-scene content, eye movements, and perceived safety in an interactive visual analytics system. Across 1,500 image-rating trials from 30 participants, fine-grained attention to damaged walls, damaged pavement, and graffiti was associated with lower safety ratings, while attention to cars and doors was associated with higher ratings; temporal views further revealed when and how viewers revisited urban cues.
Key insights
- Fine-grained disorder categories are more informative than broad semantic groupings: damaged walls had the strongest negative association with safety ratings (β = −0.442, p < .001), followed by damaged pavement (β = −0.229) and graffiti (β = −0.225).
- Context changes the interpretation of visual cues: scenes with pedestrians, vegetation, or active street elements could receive higher safety ratings even when graffiti or other disorder was present.
- Temporal gaze representations reveal early inspections, sustained attention, late visits, and revisits that are hidden by static aggregate heatmaps; gaze patterns should be treated as correlational evidence rather than causal explanations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.