Mapping sidewalk accessibility with smartphone imagery and Visual AI: a participatory approach.

Abstract

Evaluating sidewalk accessibility is conventionally a manual and time-consuming task that requires specialized personnel. While recent developments in Visual AI have paved the way for automating data analysis, the lack of sidewalk accessibility datasets remains a significant challenge. This study presents the design and validation of Sidewalk AI Scanner, a web app that enables quick, crowdsourced and low-cost sidewalk mapping. The app enables a participatory approach to data collection through imagery captured using smartphone cameras. Subsequently, dedicated algorithms automatically identify sidewalk features such as width, obstacles or pavement conditions. Though not a replacement for high-resolution sensing methods, this method leverages data crowdsourcing as a strategy to produce a highly scalable, city-level dataset of sidewalk accessibility, offering a novel perspective on the city's inclusivity; fostering community empowerment and participatory planning.This article is part of the theme issue 'Co-creating the future: participatory cities and digital governance'.