Video Compression with Diverse Contexts Using JPEG

Abstract—In the rapidly evolving digital landscape, efficient video compression is paramount for enhancing storage capacity and streaming performance without sacrificing quality. This project introduces an innovative approach to video compression, with well-established JPEG compression algorithm through a streamlined Python implementation to perform video compres- sion while maintaining impeccable visual fidelity. The standout feature of our Video Compression with Diverse Context (VC- DC) technique is its ability to significantly improve bitrate savings, achieving an impressive increase from 23.3 % to 49.03%. This breakthrough not only optimizes storage and transmission efficiency but also outperforms traditional video codecs, including advanced neural network-based methods like SOTA-HEM and DCVC-DC. Despite their complexity, these state-of-the-art tech- niques fall short in comparison to the simplicity and effectiveness of VC-DC. This project underscores the potential of combining classical compression algorithms with modern implementation strategies to push the boundaries of video compression, offering a compelling alternative to more resource-intensive solutions. Index Terms—Video Compression, Diverse Contexts, Bitrate Saving, Neural Networks