Machine Learning Based Fast QTMTT Partitioning Strategy for VVenC Encoder in Intra Coding

The newest video compression standard, Versatile Video Coding (VVC), was finalized in July 2020 by the Joint Experts Team (JVET). Its main goal is to reduce bitrate 50% over its predecessor coding High Efficiency (HEVC). Due new advanced tools and features included VVC, it actually provides high performances—for instance, Quad Tree with nested Multi-Type (QTMTT) involved partitioning block. Furthermore, VVC introduces various techniques that allow for superior performance compared HEVC, but an increase computational complexity. To tackle this complexity, a fast Unit partition algorithm based on machine learning intra configuration proposed work. formed five binary Light Gradient Boosting Machine (LightGBM) classifiers, which can directly predict most probable split mode each unit without passing through exhaustive process known as Rate Distortion Optimization (RDO). These LightGBM classifiers were offline trained large dataset; then, they embedded optimized implementation of VVenC. results our experiment show approach has good trade-offs terms time-saving efficiency. Depending preset chosen, achieves average time savings 30.21% 82.46% VVenC encoder anchor, Bjøntegaard Delta Bitrate (BDBR) 0.67% 3.01%, respectively.