Parallelize video processing

It seems to be quite a challenging process, especially when we consider that face detection is done outside of the gstreamer.

I tried to get some insights from the following links:

[1] https://discourse.gstreamer.org/t/optimizing-video-frame-processing-with-gstreamer-gpu-acceleration-and-parallel-processing/4190

[2] https://fluendo.com/blog/hype-hybrid-parallel-encoder/