Project Sidewalk

24 Mar

Thanks to our collaborators at the Universidade de São Paulo (USP) for the translations, Project Sidewalk has added its sixth language: Brazilian Portuguese! Graças aos nossos colaboradores da USP pelas traduções, o Projeto Sidewalk adicionou seu sexto idioma: o português!

We're launching RampNet, an open-source AI that helps detect curb ramps with near-human accuracy. The most amazing part? The entire project was conceived of and led by high school student John O'Meara. 🧵 A thread on what we built and why it matters.

View full thread

We're now working on incorporating RampNet into Project Sidewalk. Let's go! 🚀 Shoutout to the whole team, including John, @jaredhwang.bsky.social, Zeyu Wang, and @ismikey.bsky.social. CC @uwcse.bsky.social #Accessibility #AIforGood #OpenSource #ComputerVision #OpenData

huggingface.coprojectsidewalk/rampnet-dataset · Datasets at Hugging FaceWe’re on a journey to advance and democratize artificial intelligence through open source and open science.

Thanks to @hf.co for hosting our large dataset: huggingface.co/datasets/pro...

Curb ramps are critical for mobility—for wheelchair users, parents with strollers, travelers with luggage. But detecting them automatically in images is an open problem. We've been working on automatic curb ramp detection for over 10 years, and RampNet significantly outperforms our prior work!

github.comGitHub - ProjectSidewalk/RampNet: RampNet is a two-stage pipeline that addresses the scarcity of curb ramp detection datasets by using government location data to automatically generate over 210,000 a...RampNet is a two-stage pipeline that addresses the scarcity of curb ramp detection datasets by using government location data to automatically generate over 210,000 annotated Google Street View pan...

Try it yourself! See our GitHub for the code, dataset, and paper: github.com/ProjectSidew... 📈 State-of-the-Art Model (0.924 AP) 📚 The largest-ever curb ramp dataset (849,895 labels!) 💻 Everything is 100% open source.

4 June 2025

In collaboration with Professor Anupam Sobti and students, we are excited to announce a new pilot deployment in India! 🎉This work furthers our research in mapping and assessing sidewalk accessibility across socio-cultural and geographic contexts: makeabilitylab.cs.washington.edu/media/public...

makeabilitylab.cs.washington.eduAccessibility for Whom?We present a large-scale online survey exploring how five mobility groups—users of canes, walkers, mobility scooters, manual wheelchairs, and motorized wheelchairs—perceive sidewalk barriers and diffe...

25 Apr 2025

One map fits all? 🤔 Unfortunately, no! Our #CHI2025 study reveals significant differences in how cane, walker, scooter & wheelchair users perceive sidewalk barriers. Dataset & paper: makeabilitylab.cs.washington.edu/project/acce... This is work led by the amazing @chimichurrichu.bsky.social

Project Sidewalk reposted

Jon Froehlich

@jonfroehlich.bsky.social

23 Apr 2025

📢 HUGE NEWS! Project Sidewalk just crossed 1.1 MILLION manual sidewalk accessibility annotations across 21,000+ km of streets! 🎉 (1,163,413 labels to be exact!). To celebrate and help others build on our work, we have posted a sidewalk feature/barrier dataset to @hf.co huggingface.co/projectsidew...

26 Nov 2024

Hello! 👋 We are an open-source research project run by the @makeabilitylab.bsky.social. We have a two-pronged vision: first, to develop scalable data collection methods for sidewalk accessibility data using crowdsourcing, computer vision, and online map imagery; and second,