New version of Smart-City-Sample 21.6.1 not able to run on live rtsp source

[divdaisymuffin]

Hi @nnshah1 @xwu2git,

The new update of 21.6.1 fails to run default pipeline while its working fine with simulated videos. Please find attached logs for the same , error says rtpjitterbuffer gstrtpjitterbuffer.c:1402:gst_jitter_buffer_sink_parse_caps:<rtpjitterbuffer1> Got caps with wrong payload type (got 50, expected 105)

We need to use the latest update as it optimise the CPU consumption.

Kindly help us with this issue as soon as possible.

Thanks

[whbruce]

This issue seems to be specific to the camera, somehow the latest build is more sensitive to possible RTSP incompatibilities. Can you give us access to the camera via your "worker4" machine?

Without access to the camera it is difficult for us to debug. As a first step, let us know if the camera works with the VA Serving REST service as follows

Start the service as follows

docker run -p 8080:8080 -v /tmp:/tmp intel/video-analytics-serving

In a different console start a pipeline using curl, replacing camera-address with address of your camera. All going well, the request will display 1 as a response.

curl localhost:8080/pipelines/object_detection/person_vehicle_bike -X POST -H \
'Content-Type: application/json' -d \
'{
"source": {
  "uri": "rtsp://camera-address",
  "type": "uri"
},
"destination": {
  "metadata": {
    "type": "file",
    "path": "/tmp/detection_results.json",
    "format": "json-lines"
  }
}
}'
1

Then send us output from the service that will be displayed on the console. The last few lines should look like this

{"levelname": "INFO", "asctime": "2021-12-17 03:37:44,304", "message": "Starting Tornado Server on port: 8080", "module": "__main__"}
{"levelname": "INFO", "asctime": "2021-12-17 03:37:53,943", "message": "Creating Instance of Pipeline object_detection/person_vehicle_bike", "module": "pipeline_manager"}