Machine Learning Accessibility

Project Sidewalk is a crowdsourced effort to map street level accessibility issues around the world. The project uses Google Street View images to find out where accessibility problems are apparent on the streets.

Using Project Sidewalk you can help identify accessibility problems (such as missing curb ramps on sidewalks, obstacles on paths and uneven surfaces) by tagging them on Google Maps Street View. When you first start using Project Sidewalk you are walked through how to spot and identify accessibility problems on Street View. After completing the tutorial you can explore the streets on Street View and find and label any accessibility features that you find.

Project Sidewalk users have already mapped the accessibility of nearly half of Washington DC. Once Project Sidewalk has enough crowdsourced data it should be able to develop better computer vision and machine learning algorithms to automatically detect accessibility issues with Street View. This would enable developers to automatically create accessibility maps and routing engines for people with accessibility problems.

You might also like The History of Machine Learning and Street View, which looks at how MIT has trained computer AI's to identify the safety of city streets and neighborhoods experiencing high levels of urban development.