Face Detection Docker Container

I would like someone who has experience in OpenCV, CUDA, GStreamer and Docker. I would like for someone to build me a docker container that will allow me to stream an RTSP feed in mp4 with bounding boxes over the faces. I am only after facial detection and NOT any facial recognition. I would like the container to be able to accept an RTSP feed using openCV and Nivida GPU CUDA cores to decode and apply bounding boxes on peoples faces, once this is done I need it to stream the video in mp4.

Here is a docker container that has most of what we need, just need to add rtsp stream, bounding boxes and out a mp4 stream. https://github.com/Fizmath/Docker-opencv-GPU