For the past eight years, Google Street View cars have coasted along pretty much unaltered. That was until images of the updated version of Street View cameras surfaced online.
What is Google Street View? In short, it’s like an Internet web spider crawling around and indexing the physical world.
If you are not familiar with it, Street View is the leading technology behind Google Maps and Google Earth. It is the reason why people can view panoramic views from positions along many streets around the world.
Because of Google Street View, it is now easier for people to scout streets and destinations without having to go there. You might be wondering how Google can execute such a massive project, to think that there are over a hundred thousand, if not millions, of streets across the globe.
Simply put, Google has a fleet of vehicles fitted with special equipment and cameras to take pictures of different places. You might be familiar with one, or if you’re lucky enough, you may have spotty the goofy looking Street View car roaming the streets near your home.
A Street View vehicle is typically loaded with three major data capturing equipment: cameras, navigation sensors and a Global Positioning System (GPS), and laser range scanners. You might say that the most important part of the vehicle is the cameras, and yet, up until this point, they were the most ignored.#Google upgrades it Google Street View cameras making them more AI-focused!Click To Tweet
The last time the Google Street View cameras were updated was in 2009. The same year Google launched Smart Navigation. However, as the demand for better and more accurate views of locations surface, there is nothing left to do but to update the technology and the cameras that made Street View a must-have innovation.
Google Street View Cameras and Artificial Intelligence
Google is considered to be one of the world leaders when it comes to the study and application of artificial intelligence. So, it comes as no surprise that the company will now be incorporating AI into its Google Street View technology.
According to a report from Wired, Google will now be adopting a better camera hardware for Google Street View. The new hardware was said to be designed with more than human concerns in mind.
In the past, the Google Street View car was equipped with a much larger ball of cameras, 15 cameras to be exact, and often fell victim to image stitching errors.
The new Street View rig boasts just 7 20 megapixel lenses and sensors that are expected to provide clearer image feeds, closer shots of buildings, and street signs into Google’s image recognition algorithms. A new HD camera is used to recognize signage and business names and logos.
Apparently, better technology means better source data for Google’s AI algorithms, which will help it to expand its search and Assistant functions.
Using the most advanced hardware that will provide tons of information from different parts of the world to Google’s machine learning system, the company hopes to improve its already impressive virtual mapping system.
A decade after it was launched, Google Street View cars have already snapped over 80 billion photos in thousands of cities across 85 countries. However, Google still wanted a far better index of the world for its users. Jen Fitzpatrick, vice president of Google’s mapping division, was quoted as saying:
“People are coming to us every day with harder and deeper questions.”
It appears that people are asking tougher questions and the company intends to answer them by offering a fresher and more detailed digital model of the world. Using artificial intelligence, Fitzpatrick wants her service to handle queries that assume knowledge of what the world looks like.
According to her, questions like “What’s the name of the pink store next to the church on the corner?” require richer and deeper information to be answered.
AI to Map the World Using Images Captured by Google Street View Cameras
Google’s investment in machine learning and AI are finally paying off as algorithms can now automatically create new addresses in the company’s maps database just by locating and transcribing street names and numbers. The company said that Google Street View is the first Google product to actually use TPU, the company’s powerful custom AI chip.
The AI system learns the idiosyncrasies of regional signage by pulling from a sea of image data–it compares abbreviations to signs with explicit language and deduces the appropriate meaning. The AI was also trained to recognize business names and is said to be smart enough to ignore visual hazards such as a Bridgestone advertisement placed closely to an unaffiliated tire repair shop.
“From a machine learning perspective, everything gets better,” says Andrew Lookingbill, an engineer working on the technology.
The new Google Street View vehicles started capturing images last month and are expected to dominate the roads around the world in the coming months.
As Wired points out, with Street View, Google is doing what they do best: indexing. Only, in this case, they are indexing the world’s physical spaces. Just like with the Internet, Google’s map indexing is unrivaled–and now it’s going to get better.