Description

This presentation discusses alternatives to conventional video encoding that would enable fast and reliable video content analysis, while maintaining compliance with the mainstream video coding standards. By considering encoder motion estimation for temporal prediction, an Application-Aware Video Coding framework was developed using an unconstrained optimisation model that combines Rate-Distortion and Motion-Description-Error. It is demonstrated that the proposed optimisation framework can improve compressed domain motion description by 50%-60%, without compromising on rate distortion performance or encoder complexity.