Skip to content

docker.md

Latest commit

 

History

History
67 lines (51 loc) · 2.91 KB

docker.md

File metadata and controls

67 lines (51 loc) · 2.91 KB

VMAF Usage through Docker

Install Docker, then, from the VMAF directory, run:

docker build -t vmaf .

And to use it, just run:

docker run --rm vmaf [CLI]

Where [CLI] is one of run_vmaf.

For example, if you are under root, to run run_vmaf on a sample reference/distorted video pair under resource/yuv:

docker run --rm -v $(pwd):/files vmaf \
    yuv420p 576 324 \
    /files/src01_hrc00_576x324.yuv \
    /files/src01_hrc01_576x324.yuv \
    --out-fmt json

Note that you need to first download the test videos from vmaf_resource.

wget https://github.com/Netflix/vmaf_resource/raw/master/python/test/resource/yuv/src01_hrc00_576x324.yuv
wget https://github.com/Netflix/vmaf_resource/raw/master/python/test/resource/yuv/src01_hrc01_576x324.yuv

Use docker with GPU support

To run docker containers with GPU support you have to install the nvidia container toolkit. The above built container already has CUDA support inside and can use CUDA when the nvidia runtime is in use:

docker run --gpus all --rm -v $(pwd):/files vmaf \
    yuv420p 576 324 \
    /files/src01_hrc00_576x324.yuv \
    /files/src01_hrc01_576x324.yuv \
    --out-fmt json

While CUDA support is already fast using the vmaf tool the tool itself is heavily IO bottlenecked and a usage with ffmpeg and corresponding hardware decoders is much more efficient. To enable that we provide a separate Dockerfile that builds ffmpeg with cuda support and the vmaf filter. To build the container use:

docker build -f Dockerfile.ffmpeg -t ffmpeg_vmaf .

Usage of the container is very similar to the above examples. For example to run the vmaf filter on two mp4 files use:

wget https://ultravideo.fi/video/Beauty_3840x2160_120fps_420_8bit_HEVC_RAW.hevc

docker run --gpus all  -e NVIDIA_DRIVER_CAPABILITIES=compute,video -v $(pwd):/files ffmpeg_vmaf  \
  -y -hwaccel cuda -hwaccel_output_format cuda -i /files/Beauty_3840x2160_120fps_420_8bit_HEVC_RAW.hevc \
  -fps_mode vfr -c:a copy -c:v hevc_nvenc -b:v 2M /files/dist.mp4

docker run --gpus all  -e NVIDIA_DRIVER_CAPABILITIES=compute,video -v $(pwd):/files ffmpeg_vmaf \
    -hwaccel cuda -hwaccel_output_format cuda -i /files/Beauty_3840x2160_120fps_420_8bit_HEVC_RAW.hevc \
    -hwaccel cuda -hwaccel_output_format cuda -i /files/dist.mp4 \
    -filter_complex "[0:v]scale_cuda=format=yuv420p[ref];[1:v]scale_cuda=format=yuv420p[dist];[ref][dist]libvmaf_cuda" \
    -f null -

As you see for 420 video format we will have to convert from NV12 to 420 as seen above using scale_cuda. For other formats like yuv444p or yuv422p you can directly use the input from the decoder without conversion. -filter_complex [0:v][1:v]libvmaf_cuda