- How Important Is the Power Supply (PSU) When Building a PC?
- How to Turn a Windows Laptop Into a Desktop PC
- What Is “Binning” for Computer Components?
- 6 Things to Do With Your Old PS4, Xbox, or Other Console
- When Are Integrated Graphics Good Enough on a PC?
- Everything You Need to Know Before Buying an Xbox Series X|S
- How “Unified Memory” Speeds Up Apple’s M1 ARM Macs
- How to Expand Your Xbox Series X|S Storage
- What Is Apple’s M1 Chip for the Mac?
- How to Work Around a Broken Keyboard Key on a Windows 10 PC
Google Readies New Street View Cameras to Boost Machine Learning
Google’s car-mounted Street View cameras have been traversing the globe for a decade, offering users the opportunity to stroll down a street from the comfort of their computer screen. However, the cameras used to gather this data haven’t improved in eight years. Google is in the midst of revamping its Street View cameras, and you’ll be able to see the results when they hit the roads. It’s not only about the pretty pictures, though — this is about feeding Google’s machine learning algorithms the data they need to index the real world.
The current Street View rig has a distinctive cluster of 15 cameras inside a globe-shaped enclosure atop the car. The new design doesn’t need as many cameras — just seven of them, but each one has a 20MP resolution. There are also two separate cameras pointed forward and backward that capture HD still photos. Those cameras are present to feed clean zoomed images into Google’s algorithms to identify street signs and other objects. The final main component of the new rig is a pair of laser radar sensors that help determine distances.
Eight years is a long time to go without an imaging upgrade. Imagine if you were still using a digital camera from 2009 right now, and you can begin to fathom how much better the new Street View system is. The images show much more detail, and the colors are more vibrant and lifelike. Zooming in on current Street View images is a blurry mess, but the new ones could let you examine distant objects and really get a feel for an area.
With the improved images, Google hopes it can create a better index of where things are in the real world. Its machine learning algorithms are equipped to read street signs, store names, and even identify logos. It’s possible you could ask Google to identify a building with vague generalizations like what it’s near and what color the facade is.
This focus on having better Street View imagery dovetails nicely with several Google products. For example, the upcoming Google Lens, which was demoed at I/O 2017. With Lens, you’ll be able to get information about a business simply by pointing your phone at it, and some of that data could come from having better Street View images for Google’s algorithms to parse. And then there’s Assistant, which could answer more questions about a location based on data Google culled from Street View.
Google hasn’t said when the new camera system will start rolling out, but you’ll be able to tell when the images are updated.