Skip to navigation
Vision AgentAgent WorkflowsIndustry-Specific ExamplesSmart City BlueprintDevelopment Workflow

Deployment

View as Markdown

Customize Location

The default configurations and sample streams are designed for a specific location. To deploy the blueprint for your own location, you must complete the following steps:

  1. Create calibration files: Generate calibration files for your cameras. Refer to Calibration for detailed instructions.

  2. Update configuration files: Modify the relevant configuration files to reflect your environment settings, see details below for each deployment method.

  3. Configure your video streams: Obtain and configure the video streams for your deployment location.

The sections below describe the configuration changes required for each deployment method.

Docker Compose

All paths below are relative to the repository root.

  1. Update location information for the city in deploy/docker/industry-profiles/smartcities/vss-behavior-analytics/configs/vss-behavior-analytics-kafka-config.json

    • In this file, update the osmQueryPlace parameter to the name of your city. The value follows the OSM naming scheme. If it is set incorrectly you will see errors in the vss-behavior-analytics container logs and no behavior output in Kibana. Check with:

      docker logs --since 5m --tail 200 vss-behavior-analytics 2>&1 | grep -i "osm\|roadNetwork"

  2. Replace the calibration.json and road-network.json files in deploy/docker/industry-profiles/smartcities/smc-app/calibration/sample-data with the files generated by the Calibration process.

    • Alternatively, you can upload these via the UI or API once the deployment is up. To use the UI, go to the http://<HOST_IP>:3002 endpoint and upload the files. For the API, make a POST request to http://<HOST_IP>:7777/video-analytics-api/config/upload-file/calibration with the JSON as a payload for calibration, and to http://<HOST_IP>:7777/video-analytics-api/config/upload-file/road-network with the JSON as a payload for the road network.

  3. Delete (or move) all video files in the NVStreamer video directory, deploy/docker/vss-nvstreamer-video-dir/videos/dev-profile-alerts/, so that the sample clips are not auto-loaded on deployment. You can then either replace them with your own videos or keep the directory empty. Each file name must match the corresponding sensor name in the calibration file exactly, otherwise Analytics will not map the videos correctly.

    The directory name comes from NVSTREAMER_VIDEO_DIR in deploy/docker/industry-profiles/smartcities/overrides.env; if you changed that value, adjust the path accordingly.

  4. Redeploy the blueprint.

  5. Navigate to VST at http://<HOST_IP>:30888/vst/#/dashboard and add your RTSP streams.

    • If you copied streams to the folder mentioned above, those streams should automatically show up here. If you have video files instead and did not copy them to the above folder, you can upload them to NVStreamer via the UI at http://<HOST_IP>:31000 then use the RTSP stream link provided and add to VST.
    • Make sure that if you are manually adding streams to VST that the sensor name set matches exactly what you have in the calibration file otherwise the videos will not be mapped correctly by Analytics.
    • For additional details on how to use VST, check the VST docs page and for additional details on how to use NVStreamer, check the NVStreamer docs page.