Remember when Google Maps was just a small and simple location app? Well, you might be aware of the fact that the app grew tremendously in the past years, and has offered quite a few unique innovations as far as its market segment is concerned. As far as Maps’ most remarkable upgrades, we’ve loved these the most:

  • Local reviews in all native languages;
  • Parking difficulty icons;
  • Wheelchair-friendly pointers;
  • Full Uber integration;
  • And real-time traffic info.

Google Maps has evolved far from being a mere collection of maps from across the globe. The app is now a reliable navigation tool that everyone can use. And now, the developer team in charge of making Maps better has set out on a new and bold endeavor.

Extracting Information from Google’s Street View

The brave new project is titles ‘Attention-Based Extraction of Structured Information from Street View Imagery’ and it basically aims to take street names, as well as numbers and local businesses directly from picture off of Street View. This means that Maps will be kept up to date at all times. However, the algorithm behind it is understandably very complex, so Google are taking their time to optimize it. The project is based on deep neural networks, which allow for a proper automation of the extraction function.

The Deep Learning system will analyze data gathered by Google’s beloved Street View cars, and will identify in them particularities such as numbers and street names. Thus, mapping out a town or city becomes easier than ever. And the algorithm is also very efficient and reliable, so there is no room for mistakes. We’ll have to wait and see how the whole thing works, because as we mentioned before, Google is working hard to optimize it to the last details.