Fakultät für Technische Wissenschaften
Cisco reported in the past reports that the video data share was expected to reach 80% by the year 2023. However, due to the pandemic and recently imposed a remote work lifestyle, this figure is expected to increase even more. Except for the on-demand and conferencing services, the number of users that are generating, storing, and sharing their content usually through either social media platforms or video sharing platforms is increasing. Meanwhile from the video coding perspective, as video technologies evolve towards improved compression performance, their complexity inversely increases.
A challenge that many video service providers face is the heterogeneity of networks and display devices for streaming, as well as dealing with a wide variety of content with different encoding performance. In the past, a fixed bit rate ladder solution based on a „fitting all“ approach has been employed. However, such a content-tailored solution is highly demanding; the computational and financial cost of constructing the convex hull per video by encoding at all resolutions and quantization levels is huge. In this talk, we present a content-gnostic approach that exploits machine learning to predict the bit rate ladder with only a small number of encodes required.
Christian Timmerer (christian [dot] timmerer [at] itec [dot] aau [dot] at)