Building standalone version of Edge Video Analytics Microservice¶

The standalone version of Edge Video Analytics Microservice Overview does not contain any EII library dependencies. And as a result of this, this mode does not support the gst-udfloader plugin.

Steps to bulid the standalone image¶

Clone the Edge Video Analytics Microservice Overview repository .. code-block:: bash

$ git clone https://github.com/intel-innersource/applications.services.esh.edge-video-analytics-microservice.git $ cd applications.services.esh.edge-video-analytics-microservice
Run the build.sh script to build container image .. code-block:: bash

$ export TAG=”1.0.0-standalone” $ docker/standalone/build.sh $TAG

Steps to run the standalone image¶

Pull the container image from registry .. code-block:: bash

$ export TAG=”1.0.0-standalone” $ docker pull intel/edge_video_analytics_microservice:${TAG}
Run the image .. code-block:: bash

$ docker run -itd –rm
–privileged –device=/dev:/dev –device-cgroup-rule=’c 189:* rmw’ –device-cgroup-rule=’c 209:* rmw’ –group-add 109 -p 8080:8080 -p 8554:8554 -e ENABLE_RTSP=true -e RTSP_PORT=8554 -e ENABLE_WEBRTC=true -e WEBRTC_SIGNALING_SERVER=ws://localhost:8443 -e RUN_MODE=EVA -e DETECTION_DEVICE=CPU -e CLASSIFICATION_DEVICE=CPU intel/edge_video_analytics_microservice:${TAG}

Adding custom pipelines and models¶

Custom pipelines and models can be made available to the pipeline server by mounting them into to container image.

Prepare models

Use the model downloader available here to download new models. Point MODEL_DIR to the directory containing the new models. The following section assumes that the new models are available under $(pwd)/models.
```
$ export MODEL_DIR=$(pwd)/models
```
Prepare pipelines

Use these docs to get started with defining new pipelines. Once the new pipelines have been defined, point PIPELINE_DIR to the directory containing the new pipelines. The following section assumes that the new pipelines are available under $(pwd)/pipelines.
```
$ export PIPELINE_DIR=$(pwd)/pipelines
```
Run the image with new models and pipelines mounted into the container .. code-block:: bash

$ docker run -itd –rm
–privileged –device=/dev:/dev –device-cgroup-rule=’c 189:* rmw’ –device-cgroup-rule=’c 209:* rmw’ –group-add 109 -p 8080:8080 -p 8554:8554 -e ENABLE_RTSP=true -e RTSP_PORT=8554 -e ENABLE_WEBRTC=true -e WEBRTC_SIGNALING_SERVER=ws://localhost:8443 -e RUN_MODE=EVA -e DETECTION_DEVICE=CPU -e CLASSIFICATION_DEVICE=CPU -v ${MODEL_DIR}:/home/pipeline-server/models -v ${PIPELINE_DIR}:/home/pipeline-server/pipelines intel/edge_video_analytics_microservice:${TAG}

## Starting pipelines

We can trigger pipelines using the pipeline server’s REST endpoints, here is an example cURL command, the output is available as a RTSP stream at rtsp://:raw-html-m2r:`<host ip>`/pipeline-server

$ curl localhost:8080/pipelines/object_classification/vehicle_attributes -X POST -H \
  'Content-Type: application/json' -d \
  '{
     "source": {
        "uri": "https://github.com/intel-iot-devkit/sample-videos/blob/master/car-detection.mp4?raw=true",
        "type": "uri"
     },
     "destination": {
        "metadata": {
           "type": "file",
           "path": "/tmp/results.jsonl",
           "format": "json-lines"
        },
        "frame": {
           "type": "rtsp",
           "path": "pipeline-server"
        }
     },
     "parameters": {
        "detection-device": "CPU"
     }
  }'

## **gencamsrc** sample pipeline

Add the pipeline give below under $(pwd)/pipelines/cameras/gencam/pipeline.json .. code-block:: json

{
“type”: “GStreamer”, “template”: [ “gencamsrc serial=<serial no> name=source”, ” ! videoconvert”, ” ! video/x-raw,format=BGR”, ” ! appsink name=destination” ], “description”: “Pipeline”

}

Run the container as root, with environment var GENICAM=Matrix_Vision set and make sure /run/udev is mounted. .. code-block:: bash

$ docker run -it –rm
-u 0 –privileged –device=/dev:/dev –device-cgroup-rule=’c 189:* rmw’ –device-cgroup-rule=’c 209:* rmw’ –group-add 109 -p 8080:8080 -p 8554:8554 -e ENABLE_RTSP=true -e RTSP_PORT=8554 -e ENABLE_WEBRTC=true -e WEBRTC_SIGNALING_SERVER=ws://localhost:8443 -e RUN_MODE=EVA -e DETECTION_DEVICE=CPU -e CLASSIFICATION_DEVICE=CPU -e GENICAM=”Matrix_Vision” -e GST_DEBUG=”1,gencamsrc:2” -v ${PIPELINE_DIR}:/home/pipeline-server/pipelines -v /run/udev:/run/udev:ro intel/edge_video_analytics_microservice:${TAG}
- Trigger the pipeline using the below cURL command, the output is available as a RTSP stream at rtsp://:raw-html-m2r:`<host ip>`/pipeline-server .. code-block:
```
curl localhost:8080/pipelines/cameras/gencam -X POST -H \
 'Content-Type: application/json' -d \
 '{
    "source": {
       "element": "gencamsrc",
       "type": "gst"
    },
    "destination": {
       "metadata": {
          "type": "file",
          "path": "/tmp/results.jsonl",
          "format": "json-lines"
       },
       "frame": {
          "type": "rtsp",
          "path": "pipeline-server"
       }
    }
 }'
```