Skip to content

Devices

The fleet page. Everything the platform knows comes from these devices, so this is the first place to look when data stops arriving.

The Devices page showing fleet counts, the device list and device groups

Across the top:

Card Meaning
Total Devices Everything registered to your tenant
Online Reporting right now
Offline Registered but not reporting
Discovered Announced itself, but nobody has registered it yet

Discovered is the one people miss. A discovered device is powered on and talking, but it is not collecting anything for you until it is registered. A non-zero Discovered count is a to-do, not a status.

A rising Offline count usually means a network problem rather than several simultaneous device failures.

Filterable by status. Each device shows its name, hardware type, online state, and its latest telemetry — CPU, memory, temperature and connectivity.

Readings of -- mean the device has not reported telemetry yet. On a freshly-registered device that is normal for the first cycle; persisting is not.

Devices announce themselves over MQTT when they come online. Open the add-device dialog to see everything discovered and not yet registered, select what you want, and register it.

If nothing appears, the device is not reaching the broker — check that it is powered on and on the right network before looking anywhere else.

Groups are managed from this page, and they matter more than they look. Configuration attaches to a group, not to a device: detection settings, inspection profiles, and AI config are all set per group and delivered to every member.

A group is normally one line’s worth of cameras. Get the grouping right early — it is the unit everything else is configured against. See device groups.

Which RTSP cameras a device runs, and which single one inference runs against, is configured per device from this page — open the device’s stream settings.

To watch those cameras once configured, use Captures › Cameras. A camera that is disabled here shows as disabled there.

Device telemetry can be pushed to your own observability stack over OTLP by configuring an ingestion endpoint. This is available through the CLI (yomiro otel-endpoint).