A computer vision toolkit designed to generate pixel-level object classifications, and additionally perform other types of video filtering. Originally created to remove on-screen displays from video. - View it on GitHub