This guide will walk through the steps required to port an RTSP stream into the DNNNode for data collection using the Web UI, the Data Acquisition service, and an AI application.
1. Formatting RTSP Stream URL
As most IP cameras have login requirements for connections to the RTSP stream, you will need to format the RTSP stream URL in the following way:
For example, an Avigilon camera with the login of u:admin, pw:admin, IP:192.168.1.123 would have the following RTSP URL:
2. Validate that the DNNNode can see RTSP Stream
When powered up and connected to the network, the DNNNode will produce a Web UI client. This can be accessed by visiting the device’s IP address in a browser. The default page should be the ‘Live View’ page, and will be empty when no streams are added.
To add an RTSP stream, navigate to the ‘IP Setup’ using the panel on the left-hand side of the screen. Here, you can choose the ‘+ New Camera’ button to add a stream. Assign the stream a name, enter the formatted RTSP URL, and add a description. Click ‘Save’ to confirm.
Now, navigate to the ‘Live View’ page and verify that the stream preview has populated properly, as shown in the image below.
If you’re having troubles connecting to the RTSP stream from the RTSP page, we recommend trying to connect from your local machine via VLC first. In VLC, you can try to connect to the same formatted RTSP URL using the ‘Open Network Stream’ option. VLC will provide more thorough error messages if you’re having troubles connecting.
Once the RTSP connection has been established and verified via the Web UI, we can move onto getting the services configured.
3. Connecting RTSP Stream to Data Acquisition Container
The data acquisition container is used to pull frames from the RTSP stream that will later be used in the virtual sensor configuration tool.
Navigate to the Boulder AI Platform and login to your account. From the devices page, choose your DNNNode from the list to open the Device Details page for your Node.
Under the services panel, locate ‘BAI Data Acquisition’ and click on the ‘Gear’ icon to open its configuration page.
In the Data Acquisition configuration page, choose the proper codec, set the input type to rtsp, and enter the formatted RTSP URL in the Input Source field. Make sure that the image uploads are turned on and set the Crontab Direct Entry to pull a frame every minute. A frame should populate in the next few minutes after saving this configuration. (after configuration, feel free to turn down the frequency of uploads)
After entering the configuration, scroll down to the bottom and select ‘Save and Export Settings.’
The Data Acquisition container will now upload a from from the RTSP stream on the next Crontab trigger (the next minute in this case). You can refresh the Device Page until you see the image come up at the top left of the page after a couple minutes.
4. Connecting RTSP Stream to AI Application
Under the services panel, locate your AI application and open it's configuration page. (i.e. TrafficCounter, PeopleCounter, or Mobility Services/ObjectDetector
Now, go to the 'Sensor Configuration' page. Here, set the Input Type to rtsp and set the Input Source field to the formatted RTSP URL from before.
After entering your configuration, click the 'Save' button at the top right to save the config to your device.