Introduction

Whenever you want to add a camera to VSaaS, you also have to link it to an Edge Device.

In this article you will learn to link an edge device to a set of cameras using the “Standard AI - Generic/auto” edge template.

The idea of an edge template, is to generate an Edge Agent Configuration, using attributes from the cameras that you want to link to the device.

This is important since cameras may expose different ways to retrieve images from them and you may need different kinds of analytics for each of them, hence, the XEdge module graph will look different depending on the use case, that's why it is not ideal to create Edge Agent Configurations by hand.

The following example would only support basic object detection based sensors for a given set of cameras:

While the following example would support basic object detection, LPR, poses and tracking based sensors for a given set of cameras:

How to create an agent using the Auto Edge Agent Template

Step 0

Select the cameras you want to link to the edge devices. They must be accesible from at least 1 network interface in the device.

Step 1

Setup detection server in parent asset or in every asset if applicable. In the layout shown above, we want to process detections in a local yolo_grpc_server running at “http://localhost:50051”, hence we must setup that url in the parent asset “Lab”.

Note 1: if you have multiple cameras without a parent, you must set the detection server in at least one of the group.

Note 2: at this moment, multiple detection servers are not supported for 1 set of cameras by the “auto” template.

Note: at the moment, some workarounds are necessary in order to be able to save asset attributes. In this case, we have to mock Host* and Port* for the NVR section. This will not be needed in the future.

Step 2

Setup the url for each camera. This is the url that is going to be used to retrieve images. It can be RTSP or HTTP; if HTTP is used, it will be assumed to be a JPG snapshot source.

Note: the intention is to deprecate the Url parameter under the Camera section. Prefer to manually setup attribute with field name “url”.

Step 3

Setup the sensors for each camera. This is important because this will define the use case for that camera and, in turn, it will generate the appropriate Edge Agent Configuration (graph) for that use case.

Note: this means that, everytime that the user creates/enables or deletes/disables a sensor, they should re-generate the agent using the template.

Step 4

Generate the Edge Agent Configuration.

Note: in order to use Edge Templates, “Assisted mode” must be enabled

Note: clicking on “Save”, will instruct the xedge-agent software to execute the Edge Agent Configuration.

Feel free to inspect the generated agent at any point.

How to update an agent using the Auto Edge Agent Template

Say we want to add pose detection sensors to the example above:

Step 1

Add Pose Detection sensor

Step 2

Update existing Edge Agent Configuration

Note: to update an agent generated with a template, “Assisted mode” must be enabled.

You will then be shown the new Edge Agent Configuration. Go to the end of the configuration and click on “Show Changes” to check the specific applied updates.

In this case, all additions are related to the pose recognition module.

The last step is to save. This action will instruct the xedge-agent software to update the graph in the device to reflect on the changes.