Sending single video frame requires to use multiple RTP packets, how you will split metadata between them? Also sending video stream is not reliable and server may not be able to receive/decode all video frames sent by the app because of lost RTP packets, bandwidth limitations and other network conditions.

Would be much easier to use webrtc DataChannel and send your metadata together with FrameNumber(index) and/or timestamp. Server will receive it separately and render over received video.