Eternet Portal Documentation
Vision and Video Streams
Stream inventory
The page combines stream records from the configured stream API and Frigate when available, removes duplicate stream identifiers, and exposes stream detail pages. Frigate is an optional upstream so its outage is bounded rather than blocking all page loading.
Monitoring and review
Supported workflows include live-stream access, camera details, detection/review records, reviewed state, thumbnail/export operations, stream configuration, zones, AI configuration, face recognition, license plates, triggers, and AI service settings where provided by backend contracts.
Playback caveat
Most playback is proxied from configured camera or HLS sources. Two specifically named stream IDs are mapped to local MP4 demo fixtures in the current portal code; those clips are demonstrations, not live camera evidence.
Service boundary
The portal coordinates the UI and authenticated API calls. Video transport and detections are performed by the stream/Frigate infrastructure; connectivity, camera configuration, and permissions determine which features are available.
Operator workflow from camera to event
The Streams experience begins with the cameras and stream records returned by the configured stream service. Where Frigate responds, its camera and review representations are included; duplicate IDs are collapsed so the same camera is not shown twice. The user can open a stream detail page to inspect a camera and its available format, then review associated events and related controls. This arrangement supports a workflow from visual monitoring to event triage: identify the camera, confirm that a browser-compatible stream is available, locate a relevant detection, and inspect its thumbnail or event details. A reviewed marker is a workflow state, not a guarantee that the underlying event was classified correctly.
Video playback and detection metadata are separate channels. The stream service or Frigate may report that a camera exists while the media path is unavailable. A valid HLS or proxy URL does not prove that the camera is currently sending frames. Conversely, a live image can be available while review records are empty because no event matched the requested period or the review API is unavailable. Network, permissions, camera health, transcoding, browser codec support, and reverse-proxy configuration can each affect playback. To demonstrate live operation, verify moving frames from a real camera rather than relying on a configured stream row.
Review, export, and AI configuration
Depending on backend support, the portal exposes review operations, viewed state, export, thumbnail access, stream add/delete, recording toggles, zone changes, AI configuration, faces and recognition events, license-plate records, triggers, and AI service configuration. These are not necessarily available on every installation or camera. The connected Video/Frigate APIs determine which operations succeed, and permissions may differ by operation. Changes to stream configuration, zones, or AI services can affect downstream detections and should be tested with a controlled camera before rollout.
Face and plate information can be sensitive. A customer-facing explanation should focus on the operational purpose, access controls, retention and review policy, and the source of the detection. Do not imply that AI identification is infallible: detections can be missed or misclassified and should be reviewed where consequential. Export should be described only as an action that produces the backend-supported output, and the resulting file or link must be handled according to the customer’s data policy.
Availability and customer value
The page treats Frigate as an optional upstream so an outage there does not necessarily hold the entire workspace in a loading state. The interface may therefore show streams but no events, or events without some camera details. This partial-availability behavior is useful operationally, but it should be made visible in a demonstration. Two known stream identifiers use local MP4 fixtures; those must be identified as sample playback, not real-time surveillance. For customers, the module can centralize camera discovery, stream inspection, and event review alongside IoT monitoring. Its value depends on a healthy video backend, correctly configured cameras, and appropriate governance for recorded and identified data.