Abstract: This paper investigates the efficacy of jointly optimizing content-specific post-processing filters to adapt a human-oriented video/image codec into a codec suitable for machine vision tasks ...