I am kind of stuck. What I am trying to accomplish is this:
- read in an rtsp stream 30fps 1080p
- use jetson-inference to do object detection to determine area of the video I want to zoom in on.
- crop to detected area.
- resize cropped area to match original rtsp video size keeping aspect ratio, 30fps 1080p
- send back out new video feed as rtsp stream 30fps 1080p.
I currently have this working with a few glitches. I am trying to do this on a Jetson Xavier NX
using Python, Gstreamer, jetson-inference, OpenCV, RidgeRun (GstRtspSink). I am now trying to figure out how to get the audio part of this working. I am currently stumped on how to go about getting this to work.
Does anyone have any idea on how i should go about getting audio working in my process since OpenCV does only video?