HTTP Webhook Reference

CosmoEdge supports pushing alarm/event data to an external platform over HTTP. The push configuration is currently maintained through system interfaces, and the event payload fields come from the current event DTO and the packaged HTML interface documentation.

Configuration API

Query:

/gtw/cwai/System/QueryHttpInterfaceParam

Set:

/gtw/cwai/System/SetHttpInterfaceParam

Set request example:

{
  "switch": true,
  "url": "http://example.com/cosmo/events"
}

Query response example:

{
  "resCode": 1,
  "resMsg": [],
  "resData": {
    "enable": true,
    "switch": true,
    "url": "http://example.com/cosmo/events"
  }
}

Fields:

Field	Type	Description
switch	boolean	Whether HTTP push is enabled.
enable	boolean	Whether HTTP push is enabled; compatibility field in the query response.
url	string	Receiver URL.

Event Payload

A typical event payload:

{
  "messageId": "MSG-001",
  "devId": "DEVICE_SN",
  "taskId": "TASK_ID",
  "videoChannelId": "CHANNEL_ID",
  "channelName": "Entrance Camera",
  "timestamp": "1792147200000",
  "algorithmId": "ALG_ID",
  "algorithmCode": "helmet",
  "algorithmName": "Helmet Detection",
  "areaId": "AREA_ID",
  "areaName": "Work Zone",
  "orignalPicture": "/data/event/original.jpg",
  "fullPicture": "/data/event/full.jpg",
  "detectedPicture": "/data/event/target.jpg",
  "video": "/data/event/alarm.mp4",
  "videostructured": "/data/event/structured.json",
  "overviewFile": "/data/event/overview.json",
  "recordId": "RECORD_ID",
  "isRetryMessage": false,
  "category": "alarm",
  "targets": [
    {
      "label": "category0",
      "confidence": 0.93,
      "trackId": "TARGET_TRACK_ID",
      "box": {"x": 120, "y": 80, "width": 240, "height": 360}
    }
  ],
  "property": {}
}

For field details, see API Fields.

Note: The DTO also defines the itimestamp and files fields, but the current outbound to_json serialization does not emit them, so they are intentionally omitted from the payload above.

Detection Targets

targets lists the detection targets that triggered the event:

Field	Type	Description
label	string	Model detection label, such as category0, person, or Pedestrian
confidence	number	Detection confidence
trackId	string	Target tracking ID; may be omitted when tracking is disabled
box	object	Pixel-coordinate bounding box with x, y, width, and height

Aggregate events without individual targets may omit targets. Initial HTTP delivery and offline retries use the same structure.

Property Object

property changes with the algorithm type. The property categories that can currently be confirmed include:

Category	Description
face	Face quality, age, gender, mask, glasses, feature files, and face image.
body	Body attributes, body features, and body image.
vehicle	License plate, vehicle color, vehicle type, direction, and vehicle attributes.
behavior	Behavior count, duration, and target ID.
machineMaterial	Material/equipment status match result.
people	People flow statistics.
car	Vehicle flow statistics.
workClothesRecognition	Work-clothes recognition match result.
personCount	Person count.
countNumber	Count number.

Additional sub-objects may appear alongside the main categories:

Sub-object	Description
recognition	Face-library match result; accompanies face.
persons	Person target list; accompanies personCount.
target	Target enter/exit area time and image; optional across types.

Receiver Recommendations

• The receiver should return 2xx to indicate successful processing.
• Use messageId or recordId for idempotent processing.
• Perform lazy fetching with retry for image, video, and structured file URLs.
• Do not assume all fields in property are present; different algorithms only populate the relevant fields.
• The current implementation retains legacy field names such as orignalPicture; receivers must be tolerant of them.

Packaged HTML Reference

The repository currently retains:

data/Interface/ai-box-interface_v1.0.html

After installation, the entry point is:

web/staticfile/httpInterface.html

This HTML can be used as a historical interface reference, but open-source documentation should preferentially rely on the current DTO and Markdown references.