I have a webcam that produces mjpeg output. I would like to use this camera with detectnet to recognize if a person is coming into an area where they are not supposed to be.
I managed to cobble together an gst-launch queue that reads and parses the source and i though of sending this to an rtp input. However running detectnet rtp://@:1234 does not seem to receive anything. Any clue what I might be doing wrong?
call for the sending gst-launch is:
gst-launch-1.0 -v souphttpsrc location=http://192.168.1.1/snapshot.cgi user-id=user user-pw=password do-timestamp=true ! multipartdemux ! image/jpeg,width=1280,height=960 ! jpegdec ! videoconvert ! x264enc tune=zerolatency bitrate=90000 speed-preset=superfast ! rtph264pay ! udpsink host=jetson port=1234
the sending host is not the jetson
since i have no display on the jetson i tried to forward detectnet’s output somewhere else via rtp, so i called detectnet with:
detectnet --input-codec=h264 rtp://@:1234 rtp://destinationhost:1234
however detectnet just gives me ‘failed to capture video frame’ messages but way to few to indicate it is receiving something. According to tcpdump packets are sent.
If there is a simpler way to get the data from the webcam please do tell :)