Google Street View rival Mapillary collaborates with Amazon to read text in its 350M image database

Mapillary, the Swedish startup that wants to take on Google and others in the mapping of the world through a database of street images through crowdsourcing, is taking an interesting step in the development of its platform. The company is now working with Amazon, and specifically its Rekognition API, to detect and read text in the Mapillary database of 350 million images.

The first application resulting from the new feature will come from a large American city (that Mapillary will not name at this time), which plans to use the information that will now be "readable" from the parking signs to build a parking application.

"Parking is a super hot space and [parking information] is one of the most requested – for the data that people want to use Mapillary for," said Jan Erik Solem, CEO and co-founder of the startup based in Malmo, Sweden. He said that while parking will be the first application that he expects other cities to use, other applications will arise by matching the text in the Mapillary images and, later, they will be able to identify the exact latitude and longitude for specific locations. . "We are starting with parking signs in the US Because parking is one of the biggest problems in cities today, but text recognition will apply to many different types of objects and images, such as building facades." .

It may sound strange that a city has no control over the state of its parking lot on the street, apparently this is the case: cities, Solem said, have general analog maps that indicate parking regulations, but many of them lack digitized versions, which means that when they want to build new services (and in particular services that help them raise revenues through parking fees or parking fines) or consider whether they need to build more parking, they lack the sets of data to do it.

Mapillary cites research that says parking problems collectively cost $ 73 billion in the US UU., Presumably that includes not only the fines that people pay for staying longer, the gas they consume looking for a place, and so on; but perhaps even the income lost in areas where they stay and nobody knows.

Notably, Mapillary refers to his use of the Rekognition API as a "collaboration" with Amazon, although when I asked him about the writing, he was a little less willing to collaborate with me.

"I can not comment on what Amazon is doing or why," he said. "It would be nice to have that part of the ad, but this time it did not work." It seems that the two have been working together for a while, although that's all under NDA.

Amazon has been doing a lot in the world of mapping, however, as a provider of data to third parties and for their own services. For starters, the company is a logistic power, and part of the reason is due to its intelligent routing around the areas to discover the best way to collect, deliver and distribute the goods bought and sold in your market.

But that's just part of how mapping and location are used in Amazon. The company uses Here's maps and, at one point, it was rumored that it was one of the interested parties in buying the cartography company. Meanwhile, it has also incorporated (and acquired another mapping IP and talent in the process) a mapping API, similar to that of Google, for developers who want to build applications that use location services in any way for any Amazon device.

For now, it has mainly meant the creation of applications for your Fire tablet and the burgeoning range of Amazon Echo devices, but Amazon is also very focused on playing a role in other types of hardware, such as connected vehicles.

In January, Toyota announced that it would work with Amazon to integrate Alexa into its cars, and separately, Amazon has also been accumulating patents in the area of ​​autonomous vehicles.

In other words, there is a clear opportunity for a company like Mapillary, providing a viable alternative to Google Maps in terms of street level images, to work as a strong Amazon partner in these services, specifically in indexing and the supply of information about what is around you, and how to route you in the most efficient way from A to B.

And as a side note, as with other artificial intelligence platforms, Rekognition is also learning from each Once it is used in an application, so through its collaboration with Mapillary, it is also collecting more detailed information about street signs and what they say, and where they are used.

Mapillary itself is an interesting startup that I thought a long time ago was one that we should observe. Solem is a specialist in computer vision who sold to Apple his previous company, the facial recognition startup Polar Rose; and their last company has raised $ 24.5 million so far from companies like Sequoia, Atomico, Navinfo, BMW and Samsung, leading the way to build some interesting raids with potential customers.

One of these happens to include Here, the Amazon map provider. Others are under NDA. Together, they are now also feeding mapilla images to contribute to their broader database, although Solem said that about 80 percent of their 350 million-image content still comes from individuals, just like Waze. "Everyone seems to have the need to solve some kind of problem, or the desire to fix maps around the world," he said.

While text recognition and reading is a problem that Mapillary is addressing today, the next will be helping people capture more images, more automatically. "The next iteration of our product will help people complete the coverage," Solem said without elaborating on the details. "We are building tools to implement capture tasks."

Leave a Reply