Towards Human-AI Accessibility Mapping in India: VLM-Guided Annotations and POI-Centric Analysis in Chandigarh
A new AI tool is revolutionizing how cities map accessibility for people with disabilities.
Deep Dive
Researchers adapted the 'Project Sidewalk' platform using Vision-Language Models (VLMs) to analyze Google Street View imagery in Chandigarh, India. The AI-guided system audited 40 km of sidewalks around 230 Points of Interest (POI), identifying 1,644 specific locations needing infrastructure improvements out of 2,913 audited spots. The VLM-based mission guidance received a high utility score of 4.66/5 from annotators, proving effective for scaling urban accessibility audits.
Why It Matters
This scalable AI method can help cities worldwide quickly identify and fix accessibility barriers, improving mobility for millions.