Towards Sustainable Video Streaming: Evaluation of AI Approaches for Content Aware Encoding

With the increasing demand for video content and the associated rise in energy consumption, there is a critical need to develop sustainable solutions for digital media delivery. This paper introduces an AI-based video encoding solution designed to reduce bitrates and minimize energy consumption in online video streaming. Traditional encoding methods often result in high bitrates and increased environmental impact. Our approach leverages machine learning algorithms and AI-based approaches to analyze video content and dynamically adjust encoding parameters based on visual characteristics. Based on previous research we determined that reducing streamed video bitrates and storage needs leads to reduced energy consumption overall. Expanding on existing work, this paper evaluates the potential of different AI-approaches and ML-models and initial results demonstrate significant bitrate savings without compromising perceptual quality, leading to lower bandwidth requirements and greener streaming practices.