Netflix's VMAF combines human vision modeling with machine learning to rate video quality and encoding methods. The VMAF package includes a standalone C library and Python library with tools for training and testing custom VMAF models. Netflix also released CAMBI, a detector for banding artifacts, and offers various ways to interact with VMAF, including command-line tool vmaf, C library libvmaf, Python library, FFmpeg filter, and Dockerfile. VMAF has been adopted by the video community and specified as the standard implementation metrics tool according to the AOM common test conditions. A contribution guide is available for those interested in contributing to VMAF.
don't have tea/gui yet? download here
Copy the tea one-liner above into your terminal to install vmaf. tea will interpret the documentation and take care of any dependencies.