The Map Platform Google Maps is an integral part of everyday digital life and provides good and reliable services. But where does Google actually get the data for all the information and how do the streets, buildings, landscapes and street views get on the map? The Google Maps team took a look at the fingers and provides insights into the processes, the data sources and how the map material is kept up to date.
There is hardly a Google product that can collect as much data and provide as much information as Google Maps. The map platform has made the world a whole lot smaller, at least in digital form, and offers everyone the opportunity to visit any place in the world with a mouse click or touch. Be it only on the map or city map, from a bird’s eye view or often on street view recordings or with the help of images uploaded by users.
Building such a map platform and, above all, keeping it up to date is a mammoth task that the Google Maps team has to face every day. Because, of course, one has the right to deliver accurate and reliable data worldwide, so that every changed road layout, every new building, every street that has been renamed and much more should be displayed promptly. No dates are mentioned, but from a global perspective, this should be a very large flood of data that needs to be processed – in addition to the expansion of the platform.
The Google Maps team is using four major components to update the platform: images, data, people and AI – in that order. Much goes through the entire chain of these processes before it reaches Maps. Of course, despite all the AIs and reliable processes, many people are needed to intervene editorially. However, it is not known how many people work in this area of Google Maps.
According to the Google Maps team, images are the first and most important source of data – both satellite images and street view images. Both are the best way to determine what a place looks like, how it has changed and whether the supplied or existing data is correct with current recordings. In addition, Streetview uses a wide variety of vehicles and the Google Maps team buys satellite images from various sources.
Google Maps gets data from more than 1,000 third-party providers around the world. These are authorities, communities, NGOs, companies and many other trustworthy partners. These have many ways of providing data and are said to have just received a new tool to submit changes to city maps. Due to the widespread relevance of Google Maps, local authorities and communities naturally also have an interest in the data in Google Maps being up to date. Above you can see an example of processed submitted data.
The map platform must then be assembled from the puzzle of data and images. Google employs a “data processing team” around the world to compare images and data. How big this is, unfortunately, will not be revealed. In case of doubt, it is also this team that intervenes in the event of errors or perhaps even gets an idea of a situation themselves. We are talking about information being checked extensively and only published when there is a very high probability that it is accurate.
In the “People” section, of course, the local guides also play a role. A community of over 100 million users that keeps the map platform up to date. Unfortunately, it is not known how large the share of (wo)man power is in the changes.
Despite the team and the local guides, the flood of data and their evaluation can no longer be fully managed – at least not in detail. And so artificial intelligence has played a major role in Google Maps for years, which can combine and merge different data sources. For example, building shapes are extracted from a combination of map data, satellite images and street view recordings and are now displayed on the map with impressive accuracy.
This has actually been done for a very long time, but the algorithms of the time were more likely to ensure that many gray lumps appeared on the map that could hardly be identified as houses – you may have seen them on Google Maps. Thanks to artificial intelligence and machine learning, the systems have been given a helping hand and better shapes can be achieved. Today, the algorithms know what houses look like and can even add textures to them. Google Maps 3D will soon emerge from this.
Although we have come a long way – there are now maps of more than 220 countries and territories – our work is far from over. Different regions have different needs and their own mapmaking challenges.
Subscribe to GoogleWatchBlog on Google News | Subscribe to the GoogleWatchBlog newsletter