https://superuser.com/questions/901099/ffmpeg-apply-blur-over-face
"There's the Deface project which splits video into frames,
detects faces using OpenCV and a trained neural network,
and applies a blur to those places."
And don't worry, installing such programs is a zoo and
will test your patience. Your distro might have opencv-GPU
disabled and only opencv-CPU is available (and it runs on
one CPU core instead of all cores), and any such
proposed solution... will be miserably slow.
https://github.com/ORB-HD/deface
Since I did a project involving some of these things recently, usually
what happens is the pulling of dependencies is not perfect,
and you have to pull in some python stuff yourself. You
try to run the program, and immediately it whines
about "some include is missing".
But in terms of approaches, I expect to see words like
"neural network", "OpenCV", "CUDA" or similar, as part
of a solution. You can't do this with "Just C code".
Paul