encoding - How to mark specific frames in a video stream/insert frame metadata to later recognize it on the receiving side -

sorry rather complicated question topic. describe here endgoal clarify why marking video frames.

i need overlay infographic effects on video stream received internet/satellite (h264). effects should synced video frames appear natural. main problem infographics context depended, different each receiving client. impossible overlay them @ stream source , stream in infographics embedded.

the idea came insert myself video encoding chain, mark specific frames synch data (how?) , decode on receiving side.

to more specific = markers should "tag1" "tag2" "tag3", , receiving party show infographic tag1, tag2, tag3.

the problem is, not standard task, insert metadata frames (i know ffmpeg this, whole media file). next problem stream not media file, rather live camera, cannot preprocessed , saved.

do have idea start, or idea how task accomplished without messing encoding pipeline?