Camera Management

Cameras are the fundamental entities in Meridian VMS. Each camera represents a single RTSP video stream from an IP camera on your network. Cameras are assigned to a recording server for continuous recording and to a site in the location hierarchy for organisational purposes.

Camera management is accessible from Settings > Cameras & Sites > Cameras in the web interface.

Adding Cameras Manually

To add a camera manually:

1. Navigate to Settings > Cameras.
2. Click Add Camera.
3. Fill in the required fields:
   • Name — a human-readable label for the camera
   • RTSP URL — the primary stream URL (typically the high-resolution main stream)
   • Location — the site-level location to assign the camera to
4. Optionally fill in additional fields (see Camera Properties below).
5. Click Create.

Automatic server assignment

If you do not explicitly select a recording server, the system automatically assigns the camera to the server with the fewest cameras. This provides basic load balancing across your recording infrastructure.

After creation, the management server publishes an add_camera command via MQTT to the assigned recording server. The recorder daemon picks this up and begins streaming and recording the camera within seconds.

ONVIF Auto-Discovery

Meridian VMS can automatically discover ONVIF-compatible cameras on your network. Discovery operates through the recording servers, which scan their local network segments.

How Discovery Works

Discovery uses two complementary methods:

WS-Discovery (multicast) : Sends a WS-Discovery probe to the network broadcast address. ONVIF cameras respond with their service endpoints. This works on flat networks where multicast traffic is not filtered.

IP Range Scan : Iterates through a specified IP range and attempts an ONVIF connection on configured ports. This is more reliable across VLANs and subnets where multicast may not reach.

Configuring Scheduled Discovery

Automated discovery is configured from the Cameras settings page under the discovery section:

Enable Discovery : Master toggle for the scheduled discovery background task.

Discovery Schedule Time : The time of day (24-hour format) when scheduled discovery runs. Default: 03:00.

Discovery Interval (minutes) : How often to re-run discovery after the initial scheduled time. Default: 60 minutes.

IP Range : The network range to scan, in CIDR notation (e.g., 192.168.1.0/24) or as a hyphenated range (e.g., 192.168.1.1-192.168.1.254).

Ports : Comma-separated list of ports to probe. Default: 80,8080. Common ONVIF ports include 80, 8080, and 8899.

Credentials : Default username and password to use when probing cameras. Many cameras ship with admin/admin as default credentials.

Discovery across subnets

Since discovery runs from the recording servers, each server discovers cameras on its own network segment. For cameras on different subnets, deploy a recording server on each subnet or ensure the IP range scan configuration covers the target ranges.

Discovered Devices

When the system discovers a new camera, it creates a discovery record containing:

• IP address and port
• Brand and model (if reported by ONVIF)
• Available stream profiles (resolution, codec)
• First seen and last seen timestamps

Discovered devices appear in the Cameras settings page. From there you can:

• Add a discovered device as a camera with pre-populated RTSP URLs and ONVIF host
• Dismiss a device you do not want to add

Dismissed devices are hidden from the list but are still tracked. If a dismissed device is seen again in a subsequent scan, it remains dismissed.

Camera Properties

Each camera record stores the following fields:

Connection Settings

Field	Description	Required
name	Human-readable camera name	Yes
rtsp_url	Primary RTSP stream URL (main stream, high resolution)	Yes
rtsp_url_sub	Sub-stream RTSP URL (lower resolution, used for detection)	No
onvif_host	ONVIF service hostname or IP (e.g., 192.168.1.100)	No
username	Camera credentials username	No
password	Camera credentials password	No

Sub-stream for detection

When a sub-stream URL is configured, AI object detection uses the lower-resolution sub-stream instead of the main stream. This significantly reduces GPU memory usage and inference time without meaningfully affecting detection accuracy. The main stream is always used for recording.

Assignment

Field	Description
location_id	The site-level location this camera belongs to
server_id	The recording server responsible for this camera

Stream Metadata

These fields are populated automatically by the recording server when it connects to the camera stream:

Field	Description
stream_width	Video width in pixels
stream_height	Video height in pixels
stream_fps	Frames per second
bitrate_kbps	Measured bitrate in kilobits per second

Retention

Field	Description
retention_days	Per-camera retention override. When set, this camera’s recordings are retained for the specified number of days instead of the global default.

Enabling and Disabling Cameras

Each camera has an enabled toggle:

• Enabled — the camera is active. The recording server streams, records, and runs detection on it.
• Disabled — the recording server stops all processes for this camera. Existing recordings are preserved.

When you disable a camera, the management server sends a command to the recording server, which gracefully stops recording and releases the stream for that camera.

When you re-enable a camera, an add_camera command is sent and recording resumes.

Camera Status

The camera status is reported by the recording server through its heartbeat every 30 seconds:

Status	Meaning
recording	The recording engine is actively capturing and the stream is healthy
error	The camera stream is unreachable or the recording process has failed. The system will attempt to restart automatically.
offline	The camera is not being processed (disabled or server offline)

Camera status is visible in the Cameras list page, in the location tree sidebar, and on the camera cards in the Live View.

Detection Zones

Detection zones allow you to define specific areas within a camera’s field of view where object detection should trigger events. Detections outside of defined zones can be ignored, reducing false positives from areas you are not interested in.

Creating a Zone

1. Navigate to Settings > Cameras and select a camera.
2. Open the Detection Zones panel.
3. Click Add Zone.
4. Draw a polygon on the camera snapshot by clicking to place points.
5. Give the zone a name and optionally set a colour for visual identification.
6. Click Save.

Zone Properties

Property	Description
name	A descriptive name for the zone (e.g., “Entrance”, “Parking Lot”)
points	List of polygon vertices as normalised coordinates (0.0 to 1.0 for both x and y)
enabled	Toggle to activate or deactivate the zone without deleting it
color	Hex colour code for rendering the zone overlay (e.g., #ff0000)
classes	Per-zone class filter (see below)

Per-Zone Class Filters

Each detection zone can have its own list of object classes that should trigger events. This allows fine-grained control:

• Zone “Entrance”: detect person only
• Zone “Parking Lot”: detect car, truck, bus, motorcycle
• Zone “Perimeter”: detect person, dog

When a zone’s class filter is set to null (empty), all classes from the server-level detection configuration are active in that zone.

Use case: reducing false positives

A camera overlooking a road and a building entrance can have two zones. The road zone filters for vehicles only, while the entrance zone filters for people only. This prevents cars on the road from generating person alerts and vice versa.

How Zones Are Applied

The detection engine evaluates each detection against all enabled zones for that camera. A detection is recorded as an event only if:

1. The detection’s bounding box centre falls within an enabled zone, and
2. The detected class is in the zone’s class filter (or the filter is empty, meaning all classes are accepted).

If no zones are defined for a camera, all detections within the frame are accepted.

Time Synchronisation

Camera clock drift can cause timestamps on ONVIF events and motion detection to be inaccurate. Meridian VMS includes a time synchronisation service that syncs camera clocks to the management server’s time via ONVIF.

How It Works

1. The management server groups all ONVIF-enabled cameras by their assigned recording server.
2. For each group, it sends a sync request to the recording server.
3. The recording server connects to each camera via ONVIF and sets the camera’s system time.
4. Results are reported back, categorising each camera as synced, already OK, or failed.

Configuration

Time sync is configured from the Cameras settings page:

Sync camera time daily : Enable or disable the daily automatic sync. When enabled, the sync runs once per day.

Sync Now : Trigger an immediate sync of all cameras. Useful after adding new cameras or after a power outage.

Sync Results

After a sync completes, the UI shows:

• Number of cameras synced (clock was adjusted)
• Number already OK (clock was within tolerance)
• Number failed (camera unreachable or ONVIF error)
• Total cameras processed

ONVIF host required

Time synchronisation only works for cameras that have an onvif_host configured. Cameras without ONVIF credentials are skipped.

Deleting a Camera

When you delete a camera:

1. The management server contacts the recording server to delete recording files from disk (best-effort).
2. All associated records are removed from the database: events, detection zones, recordings, and the camera record itself.
3. A remove_camera MQTT command is sent to the recording server to stop all processes for that camera.

Irreversible

Deleting a camera permanently removes all its recordings, events, and detection zones. This action cannot be undone.

API Reference

Method	Endpoint	Description
GET	/api/cameras	List all cameras (supports location_id filter)
POST	/api/cameras	Create a new camera
GET	/api/cameras/{id}	Get a single camera
PATCH	/api/cameras/{id}	Update camera properties
DELETE	/api/cameras/{id}	Delete a camera and all associated data
GET	/api/cameras/{id}/heatmap	Get aggregated heatmap grid
POST	/api/cameras/{id}/heatmap/generate	Generate heatmap from historical events