The growth of chart data is wonderful. Facebook users are posting 350 million photos everyday. Snapchat gets over 400 million photos a day. Instagram sees more than 27,800 photos every minute. Google maps has over 20.5k TB of data. Whatsapp users also share 400 million photos every day.
Big Data experts are needed to sort, store, meta data and other processes of these images. The scientists at UC Berkeley have simplified this process by using a method called ‘average the data’. Three scientists namely, Jun-Yang Zhu, Yong Jae Lee and Alexie Effros are working on this software project to average images. The software that they have developed, AverageExplorer has capability to create average image from cluster of images. They presented the research paper of this project at SIGGRAPH conference in Vancouver, Canada this year.
Chain of related images is easy to analyse and sorted. For example graduation ceremony or bride-grooms weddings. These kind of images can be sorted easily. Online shopping, sport analysis, media analysis can be the target by AverageExplorer. The lead researcher, Yong Jae Lee said that, the computer vision systems need to deduce the key features in order to improve the service offered by this product. For example, a lot of data input is needed for face gratitude system.
Jason Salavon, an artist has created average images using hundreds of pictures based on Chirstmas. He has posted his work on pictures taken in Chicago on the website. The scientists have automated this process of sorting and alignment of every picture.
We can create average image of average images, it could be inception like image, it will look like images within an image.