Leveraging Street Level Imagery for Urban Planning

Significance We show that socioeconomic attributes such as income, race, education, and voting patterns can be inferred from cars detected in Google Street View images using deep learning. Our model works by discovering associations between cars and people. For example, if the number of sedans in a city is higher than the number of pickup trucks, that city is likely to vote for a Democrat in the next presidential election (88% chance); if not, then the city is likely to vote for a Republican (82% chance).

Significance Attaching sensors to crowd-sourced vehicles could provide a cheap and accurate way to monitor air pollution, road quality, and other aspects of a city’s health. But in order for so-called drive-by sensing to be practically useful, the sensor-equipped vehicle fleet needs to have large “sensing power”—that is, it needs to cover a large fraction of a city’s area during a given reference period. Here, we provide an analytic description of the sensing power of taxi fleets, which agrees with empirical data from nine major cities. Our results show taxis’ sensing power is unexpectedly large—in Manhattan; just 10 random taxis cover one-third of street segments daily, which certifies that drive-by sensing can be readily implemented in the real world.