Mapillary, a startup on a mission to “fix” the world’s maps, partnered with Amazon’s machine vision program, Rekognition, to build street-level maps.
Starting with a parking app, the Swedish company plans to build a platform that rivals Google’s Street View. But the partnership raises questions about Amazon’s mapping ambitions that Mapillary can’t seem to answer.
Americans waste $73B a year looking for parking spots. Like any good Swedish designer, Mapillary thinks that’s a bit excessive.
Mapillary plans to use its database of more than 350m crowdsourced street-level photos to “read” street signs and develop smart parking maps, replacing the analog parking systems that still exist in many cities.
The company, which has raised $24.5m, already secured a contract from a “large [unnamed] US city” to build parking maps.
Mapillary has also stated it will use Amazon’s Rekognition API — which is infamous for its secret use as a law enforcement tool — to detect and read text on street signs.
But when TechCrunch pressed for partnership details, a spokesman said, “I can’t comment on what Amazon is doing, or why. It would be nice to have that part of the announcement, but it ended up not working out this time.” And Bezos slowly withdrew the knife held against his throat (jk)…
Amazon’s evasiveness is noteworthy. The company recently launched its own mapping API, and if they are about to tap into a 350m-picture crowdsourced image database, you can bet they’re gonna use it.