On December 3, 2020, Fraunhofer FOKUS' business unit Future Applications and Media (FAME) will be participating at the Mile High Video workshop. At 10:30 am (MT), our expert, Daniel Silhavy, will be providing a live presentation (followed by a Q&A session) - discussing the latest developments in improving video streaming with machine learning methods. 

Abstract

Nowadays, video streaming content is responsible for immense amounts of data on the Internet. In order to support the Internet's limited transmission capacities, videos are compressed during encoding. However, this often results in a loss of quality – which poses particular challenges for live streaming where there are limited possibilities for efficient caching. 

This is where “FAMIUM Deep Encode” comes in – it analyzes the video and collects playback metrics for each individual scene to predict and automatically adjust upcoming encoding settings. Our solution utilizes artificial intelligence methods to enable per-title encoding for Video on Demand and live streaming. In comparison to traditional video encoding methods, this approach results in a reduction of up to 30 percent in average storage and transfer volumes. This reduction leads to significant cost savings in the long-run, and improves the overall quality of experience for the end user.