Milestone XProtect Integration#
This guide explains how to run the Loitering Detection application on cameras managed by the Milestone XProtect VMS. Milestone’s AI Bridge provides the API, while the VMS Adapter Plugin connects XProtect to the application, handling app registration and translating metadata at runtime.
When you finish this guide, you will have:
Loitering Detection registered in XProtect as an IVA (Intellegent Analytics Video) app.
Detection boxes and the loitering zone drawn on the camera video in Smart Client.
An analytics event for each loitering person, raising an alarm in Smart Client’s Alarm Manager.
Concepts#
Milestone XProtect#
XProtect is Milestone’s video management software (VMS). It runs on Windows and is made of several services and clients:
Component |
Role in this integration |
|---|---|
Management Server |
Holds the configuration: cameras, users, rules, alarms. AI Bridge attaches to it. |
Recording Server |
Connects to the cameras and records their video. |
Event Server |
Receives analytics events and raises alarms from them. |
Management Client |
The administrator’s tool. You use it to subscribe the analytics topics for each camera and define alarms. |
Smart Client |
The operator’s tool. It shows live and recorded video with detection overlays, and the Alarm Manager. |
Milestone AI Bridge#
AI Bridge is a free Milestone component, delivered as Linux containers, that lets third-party Intelligent Video Analytics (IVA) apps work with XProtect. It runs on a Linux host, connects to the XProtect Management Server, and gives an IVA app:
A GraphQL API to discover cameras and to register the app.
An RTSP restream of each camera, so the app never talks to the cameras directly.
Topics the app posts results to. A metadata topic carries detection boxes into the video stream. An event topic carries analytics events to the Event Server.
Registering an app makes it appear in Management Client, where an administrator turns on its topics for each camera.
Loitering Detection#
Loitering Detection is an IVA application from the Metro Vision AI App Recipe. It runs a DL Streamer Pipeline Server (DLSPS) pipeline that detects and tracks people, and measures how long each person stays in a zone. The pipeline publishes its results on an MQTT broker. Along with detections, it also publishes one loitering status row for each person who stays within the defined zone longer than a threshold.
Where VAP fits#
Neither AI Bridge nor Loitering Detection knows about the other. VAP connects them:
Registers Loitering Detection with AI Bridge as an IVA app.
Discovers the XProtect cameras through AI Bridge.
Starts a Loitering Detection pipeline on a camera’s RTSP restream, with the zone and threshold you set. VAP passes the RTSP stream through to the pipeline, isolating Loitering Detection from AI Bridge.
Reads the pipeline’s results from MQTT, and posts them to AI Bridge: detection boxes and the zone to the metadata topic, loitering occurrences to the event topic.
Architecture#
VAP connects Milestone XProtect through AI Bridge to Loitering Detection, passing through the RTSP stream and forwarding analytics results back to AI Bridge.
flowchart LR
subgraph WIN["Windows host: Milestone XProtect"]
MS["Management Server"]
RS["Recording Server"]
ES["Event Server"]
MC["Management Client"]
SC["Smart Client"]
end
CAM["IP camera"] --> RS
subgraph AIBH["Linux host: AI Bridge"]
AIB["AI Bridge<br/>GraphQL :4000<br/>RTSP :8554<br/>topic sinks :8382"]
end
subgraph VAPH["Linux host: VMS Adapter Plugin"]
VAP["VMS Adapter Plugin<br/>backend :8080, UI :3443"]
end
subgraph LDH["Linux host: Loitering Detection"]
DLS["DL Streamer<br/>Pipeline Server"]
MQ["MQTT broker :1883"]
FL["mqtt-table-flattener"]
end
AIB <--> MS
RS --> AIB
AIB -- "metadata, events" --> RS
AIB -- "events" --> ES
VAP -- "register, discover cameras" --> AIB
VAP -- "start / stop pipeline" --> DLS
AIB -- "RTSP" --> VAP
VAP -- "RTSP passthrough" --> DLS
DLS -- "detections" --> MQ
MQ --> FL
FL -- "loitering rows" --> MQ
MQ --> VAP
VAP -- "metadata frames, analytics events" --> AIB
AI Bridge, VAP and Loitering Detection can share one Linux host.
sequenceDiagram
autonumber
participant Admin as Administrator
participant MC as Management Client
participant AIB as AI Bridge
participant VAP as VAP
participant DLS as Loitering Detection
participant SC as Smart Client
VAP->>AIB: register "Loitering Detection" (metadata + event topics)
AIB->>MC: app appears in Management Client
VAP->>AIB: discover cameras
Admin->>MC: tick the metadata topic on the camera, define an alarm
Admin->>VAP: start Loitering Detection on a camera
VAP->>DLS: start pipeline (VAP RTSP passthrough, zone, MQTT topic)
loop every frame
DLS->>VAP: detections (MQTT)
VAP->>AIB: ONVIF metadata frame (boxes + zone)
AIB->>SC: overlay on live video
end
loop every event
DLS->>VAP: loitering row (MQTT)
VAP->>AIB: analytics event
AIB->>SC: alarm in Alarm Manager
end
What you get in XProtect#
Result |
AI Bridge topic |
Where you see it |
|---|---|---|
Detection boxes and the loitering zone |
|
Smart Client live and playback video |
Loitering events |
|
Alarm Manager, rules, event log |
See Reference and Troubleshooting for what is not yet supported.
Tested versions#
Component |
Version |
|---|---|
Milestone XProtect |
Professional+ 2026 R1 |
Milestone AI Bridge |
3.0.1, 4.0.0 |
VMS Adapter Plugin |
2026.3 |
Loitering Detection |
2026.3 |
Cameras |
AXIS P3265-LVE (1920×1080, H.264) |
How to use this guide#
Follow the pages in order:
Set Up XProtect and AI Bridge: hosts, ports, XProtect and AI Bridge configuration, and how to check that AI Bridge sees your cameras.
Run Loitering Detection with XProtect: bring up Loitering Detection and VAP, turn on the topics in Management Client, start a pipeline, and see the overlays in Smart Client.
Loitering Events and Alarms: turn loitering occurrences into XProtect alarms.
Reference and Troubleshooting: configuration reference, how registration works, capabilities, limitations, and troubleshooting.