FogRecognitionAPI.md

FOG Recognition API

How to connect to CallbackAPI

First, you need to set an URLs of your backend where callbacks request will be sent. This action is available in the Sources section, you need to choose your video source then set up two URL:

1. Person meta info callback URL
2. Person visit callback URL

Event types

PersonChanged - will be sent to the Person meta info callback URL.

{
  "personId": "c15290a1-3e2e-457d-8d2f-21bd157615d0" (uuid),
  "sex": "male" (string | [male, female]),
  "age": 42 (int),
  "cameraId": "8c62b232-5cde-4d48-8a2c-67322671d12f" (uuid)
}

Field descriptions

Title	Description
PersonId	Unique person's identifier
Sex	Person's gender (Male, Female)
Age	Person's age
CameraId	Video source identifier

RecognitionEvent - will be sent to the Person visit callback URL.

{
  "cameraId": "afe78665-4398-4c5e-b2fa-31bf19ef1b4e" (uuid),
  "id": "4aec96fd-acdf-4f05-a904-87cc11edf9a4" (uuid),
  "projectId": "21b224ae-2759-434b-be96-1f82146b0a0c" (uuid),
  "taskId": "0fd6e9d5-8b51-498c-ad4a-615c372410b6" (uuid),
  "trackletId": "245bc02b-0ad6-445b-bea9-5f8125bdcb44" (uuid),
  "personId": "ac66866c-cb65-4a5a-be62-3fe3125e88d3" (uuid),
  "nearestPersons": ["eb849ce0-2e08-4042-bce1-ee70afec3413", "c38b06f8-9cfa-46cf-a928-0a8bb3000aee, "263d523a-4d4f-4ae0-8ecd-cbfc0e552f38" (uuid)],
  "isPersonified": true (bool),
  "avatarHash": -5467464835834843 (int, null),
  "tracks": [
      {
         "bbox":[
            {
               "x":1585 (int),
               "y":267 (int)
            },
            {
               "x":1641,
               "y":323
            }
         ],
         "timestamp":1565955460073 (int),
         "pts":7335336386.0 (double)
      },
      {
         "bbox":[
            {
               "x":1449,
               "y":327
            },
            {
               "x":1518,
               "y":396
            }
         ],
         "timestamp":1565955460113,
         "pts":7335339986.0
      },
      {
         "bbox":[
            {
               "x":1454,
               "y":335
            },
            {
               "x":1523,
               "y":403
            }
         ],
         "timestamp":1565955460193,
         "pts":7335347186.0
      }
   ],
  "fromTs": 1565955458713 (int),
  "toTs": 1565955460273 (int),
  "bestFaceScore": 0.7 (double | [0...1])
}

Field descriptions

Title	Description
CameraId	Video source identifier
Id	Event identifier
ProjectId	Project identifier
TaskId	Video stream identifier
TrackletId	Person's track identifier between video frames
PersonId	Person's identifier. Person will be created based on face tracklets
NearestPersons	List of similar person's identifiers
IsPersonified	Flag, shows the state of tracklet identification
AvatarHash	Hash of the best face image
Tracks	List of the person detection events. Field contains:
	Bbox - List of the face coordinates (x,y pixels) on video frame.
	Timestamp - time of detection event
	Pts - Presentation Time Stamp of frame
FromTs	Event start time
ToTs	Event end time
BestFaceScore	Face image quality score (from 0 to 1)

Faceter Fog integration

RecognitionEvent processing

• When you need to get best face image (avatar) your backend should return success response with HTTP code 200 (OK):

	"avatarCallbackUrl": "http://localhost/avatar"

Title	Description
AvatarCallbackUrl	URL which will listen to HTTP POST method of best face image

The best face image will be sent to avatarCallbackUrl with query-parameters avatarHash, personId, projectId, which contains in RecognitionEvent. Your backend should return HTTP code 204 (NoContent).

• If you do not need the best face image, it should return the HTTP code 204 (NoContent).

PersonChanged processing

• Your backend should return the HTTP code 204 (NoContent).