Automatic view tracking in 360º multimedia using xAPI
360-degree video can consume up to six times the bandwidth of a regular video by delivering the entire frames instead of just the current viewport, introducing an additional difficulty to the delivery of this kind of multimedia. Many authors address this challenge by narrowing the delivered viewport using knowledge of where the user is likely to look at. To address this, we propose an automatic view tracking system based in xAPI to collect the data required to create the knowledge required to decide the viewport that is going to be delivered. We present a use case of an interactive 360º video documentary around the migratory crisis in Greece. In this case, omnidirectional content is recorded using a 6-camera array, rendered in an equirectangular projection and played later by an HTML5 web application. Interactive hotspots are placed on specific coordinates of the space-time of the production, introducing a connection of the viewer with the story by playing additional multimedia content. The current view and other usage data are recorded and permit us to obtain metrics on the user behavior, like most watched areas, with the goal to obtain that required knowledge.
It is a follow-up of the article submitted to that conference, giving more insight on how this approach can be used to gather knowledge on the use of 360º multimedia so view-adaptive streaming can be performed.