HIGHLIGHTS
- The RAISR technique was first shown in November 2016
- RAISR uses machine learning to retain original shapes and details
- Google has been applying RAISR to more than 1 billion images per week
Artificial intelligence and machine learning to a significant leap in 2016 with Google launching Assistant for its Pixel and Home products. The company in November also showed off a machine learning technique called RAISR that could be used to enhance low-res and blurry images. 2017 will be no less for Google, which looks to improve the technology further. The RAISR technique is now being used in Google+ which will display high resolution photos using only a third of the bandwidth.
Google in a blog post said that its social networking platform, Google+, will now use the new Rapid and Accurate Image Super-Resolution (RAISR) software which, through machine learning, will allow users to see large high-resolution photos while using 75 percent less bandwidth. This will significantly help those who find themselves with spotty Internet connection and high data costs.
Essentially, RAISR uses machine learning to retain original shapes and details more accurately so as to provide clearer images. As an example, a 100KB 1000×1500 pixels image is replaced with a 25KB file having retained the original resolution through RAISR.
Google said that it has been applying RAISR “to more than 1 billion images per week, reducing these (Android) users’ total bandwidth by about a third.” The company plans to roll out the the technology more broadly in the coming weeks so users.
Sourse: gadgets.ndtv.com
