Quantifying Street-Level Accessibility: Algorithm meets Crowd-Sourcing

C ollecting and visualizing accessibility information are always challenging problems within the realm of Human-Computer Interaction (HCI). A team of researchers at the University of Maryland have proposed innovative approaches that combine crowdsourcing, computer vision and machine learning to detect and visualize how accessible a particular street or sidewalk is. Tohme is one such approach for finding curb ramps remotely in Google Street View scenes. Based on experiments conducted by the team, the system demonstrates a performance comparable to a conventional manual approach but with a dramatic reduction in time and cost.