In the Settings tab you can manage various settings aspects such as view the hubkit’s versions, backup or restore, set the base property profiles, deploy image inference models or view the disk usage space. If you have the enterprise version, you can also connect to your Gravio Coordinator here and set up and manage the blockchain functionalities.
HubKit Information
Here you can see the various aspects of the HubKit that you are connected to. This information is also important when reporting bugs so the Asteria team know which version you are operating.
At the bottom you can also set the mail server parameters that will be used to send you notifications via e-mail. These notifications can be info about low disk space or issues with the hubkit.
[Gravio 4.5 or higher] – If periodic communication from the sensor of a Zigbee device (about 60 minutes) is not received, and if data is not received for a period of time in that state (90 minutes), the system will notify the email address set in the notification settings that communication from the device has been lost. CO2 is sent out when no signal is received for 1 minute.
Backup & Restore
You can backup device configuration settings to your Gravio Cloud account, whether it’s the Asteria hosted one or the one hosted on your own local Gravio Coordinator. From there you can restore these settings back into your home folder on your Gravio account.
Base Property Profile
These are re-usable settings that will not be included in actions or backups. Examples for such settings are authentication tokens, passwords, login details, secret keys, mail server settings etc. This allows you to re-use them in your actions where needed.
Feature Package
A feature package is a set of configurations including areas, layers, triggers, actions that you can export as a Zip file to be imported on either your cloud account, or, if you have the Enterprise edition of Gravio, to your coordinator for distribution to your edge nodes.
Attention: that we recommend that you put all confidential data such as tokens, usernames, passwords for third party services into a base property profile (see above) so those details are not exported included in the package, if you tick the appropriate checkbox during the export process.
Image Inference Models
In this section, you can deploy the available image inference models to your local HubKit for use. There are two ways of deploying computer vision models:
1. Locally, via Gravio Studio (which is available for all versions including Free, Basic and Standard)
2. Via the Gravio Coordinator backend (which is only available for Enterprise licenses)
There are also a few standard models that come from our Gravio Cloud. They include a pre-trained people counting model.
To upload a new computer vision model locally via Gravio Studio, press the “Create” button, select your .tflite file package containing and upload it.
Alternatively, click the “Import” button and import a previously exported zip file.
Once you click the “Create” button, you will see the following dialogue box:
Select the Recognition Task you would like to conduct. You can either focus on detecting objects or to classify an image. For Image Classification you will need to use TensorFlow Lite.
For TensorFlow prepare the following files:
| File |
|---|
| TensorFlow meta file |
| TensorFlow pb file |
In the dialogue box:
| Form Field | Explanation |
|---|---|
| Model Package Name | Use alphanumeric characters to describe your package. This name will appear as “Sensor”. |
| TensorFlow metafile | Select the prepared file |
| TensorFlow pb file | Select prepared file |
| Method | Select Count to output a number or Group By to return a string |
| Output format | Select JSON to output detailed information, or select Value to output only the value. |
| Include detected values | Select this checkbox to include detected values |
| confidence threshold | Set between 0.0 and 1.0 to define at which threshold level the “sensor” should be fired |
*If you pick the TensorFlow Lite method, you will see the following screen instead. *
In this case prepare the following files:
| File |
|---|
Lite.tflite file |
LiteEdge.tflite file |
labels file |
In the dialogue box:
| Form Field | Explanation |
|---|---|
| Model Package Name | Use alphanumeric characters to describe your package. This name will appear as “Sensor”. |
| Lite.tflite file | Select the prepared file |
| LiteEdge.tflite file | Select prepared file |
| Labels File | The file containing the labels to be detected |
| Method | Select Count to output a number or Group By to return a string (Note: this cannot be specified if “Image Classification” is selected. |
| Output format | Select JSON to output detailed information, or select Value (or Name of you selected “Image Classification) to output only the value. |
| Include detected values | Select this checkbox to include detected values |
| confidence threshold | Set between 0.0 and 1.0 to define at which threshold level the “sensor” should be fired |
Disk Management
In Disk Management you can view the health status of your disk and set the various parameters to keep the disk space used optimally.
Need more help with this?
Join our slack community for help