The following article is based on my own interpretation of the said events and/ or publicly available information. Any material borrowed from published and unpublished sources has been appropriately referenced. I will bear the sole responsibility for anything that is found to have been copied or misappropriated or misrepresented in the following post.
Vivek Kumar Jha, MBA 2016-18, Vinod Gupta School of Management, IIT Kharagpur
With our love soaring at all time high for Social Media, uploading an uncountable number of selfies, usfies, and those trip albums every day has become a habit for most of us. According to BusinessInsider, more than 1.8 billion images are uploaded everyday on Social media platforms like Facebook, Flicker, Instagram and Whatsapp. But, we hardly think about the backend processes that these Social Media companies undergo to make the experience pleasing. With increasing resolution of the cameras, the images size up and to view them high bandwidth of the internet is utilized. This tends to be a costly affair for people residing in countries where internet connection is spotty and comes with high cost. Google has found the solution to this problem in the lap of Machine Learning and has come up with RAISR.
RAISR or Rapid and Accurate Image Super Resolution is a technology that employs Machine Learning to generate high-quality versions of low-resolution images. So, an image of 100 KB can be efficiently reduced to 25 KB without compromising the image quality. The underlying process involves the learning and training on a pair of a high and a low-quality image. It results in the creation of filters that are then applied to each pixel of the low-quality image, adding more details to it. Now, even though there have been other technologies that work in the same direction, the benefit that Google boasts about RAISR is that it is 10 to 100 times faster and reduces the bandwidth usage by more than 75%.
Google has already started testing it on Google+ platform and a set of Android devices by implementing this technology on 1 billion images every week. Google certainly seems excited about how this technology is going to evolve, making user experience only better.