How to change the Input Video Source#

Typically, a pipeline is started with a cURL request with JSON payload containing source, destination and parameters. For example, the following cURL request start an AI pipeline on a file inferencing on weld porosity classification model.

     curl http://<HOST_IP>:8080/pipelines/user_defined_pipelines/<pipeline_name> -X POST -H 'Content-Type: application/json' -d '{
        "source": {
           "uri": "file:///home/pipeline-server/resources/videos/welding.avi",
           "type": "uri"
        },
        "parameters": {
           "classification-properties": {
              "model": "/home/pipeline-server/resources/models/weld-porosity/deployment/Classification/model/model.xml",
              "device": "CPU"
           }
        }
     }'

To change the input video source for the pipeline, refer to the following table:

Video Source

Source Section of the cURL Request

Remark

File

 

”source”: {
  “uri”: “file://path”,
  “type”: “uri”
}

RTSP

 

”source”: {
  “uri”: “rtsp://url”,
  “type”: “uri”
}

In the values.yaml file inside the helm folder in the repository for helm based deployments, or in the .env file at the root of the repository for docker compose based deployment, update RTSP_CAMERA_IP to the IP of the machine where the RTSP stream is coming from:
RTSP_CAMERA_IP=<IP_where_RTSP_stream_is_originating_from>

Web Camera

 

”source”: {
  “device”: “/dev/video0”,
  “type”: “webcam”
}

The pipeline in pipeline-server-config.json in the helm chart needs to be changed as follows: 

”pipeline”: “v4l2src device=/dev/video0 name=source ! video/x-raw,format=YUY2 ! videoconvert ! video/x-raw,format=RGB ! gvaclassify name=classification model-instance-id=inst0 ! queue ! gvawatermark ! gvafpscounter ! appsink name=destination”,
`

This Page