Send raw tracking data

OMNI Agent allows sending raw tracking data of faces and silhouettes to external systems.

Supported protocols:

• MQTT
• Data will be sent to MQTT topic <default_mqtt_topic> \ <raw data type>. For example, events\body_tracking

Raw data specification

Body detection and tracking (body_tracking)

{
    // information about videostream source
    "stream_name":"rtsp://user:password@your/stream/path",
    "stream_title":"your_title",
    "stream_uuid":"3325d094-2e01-4199-931d-876200ed5b68",
    "__positional_camera_index":0,
    "camera_status":2,
    "camera_status_msg":"CAMERA_OK",
    "camera_type":"native",

    // frame information
    "timestamp_millisec":1720777422166,
    "camera_frame_data": {
        "frame_index":5457,
        "timestamp_microsec":1720777422166462,
        "timestamp_millisec":1720777422166
    },
    "image": {
        "color_space":"BGR",
        "dtype":"uint8_t",
        "format":"NDARRAY",
        "shape":[1080,1920,3]
    },
    "is_benchmark_frame":false,

    // information on tracked objects in the frame
    "objects":[
        {
            "bbox": [0.3569, 0.5603, 0.5064, 1.0],
            "class":"body",
            "confidence":0.6813,
            "id":2,
            "object_id":"4165cad2-49b6-40ce-a4f5-44eb363a4080"
        }
    ]
}

Body loss (body_lost)

{
    // information about videostream source
    "stream_name":"rtsp://user:password@your/stream/path",
    "stream_title":"your_title",
    "stream_uuid":"3325d094-2e01-4199-931d-876200ed5b68",
    "__positional_camera_index":0,
    "camera_status":2,
    "camera_status_msg":"CAMERA_OK",
    "camera_type":"native",

    // frame information
    "timestamp_millisec":1720777422166,
    "camera_frame_data": {
        "frame_index":5457,
        "timestamp_microsec":1720777422166462,
        "timestamp_millisec":1720777422166
    },
    "image": {
        "color_space":"BGR",
        "dtype":"uint8_t",
        "format":"NDARRAY",
        "shape":[1080,1920,3]
    },
    "is_benchmark_frame":false,

    // information on lost objects
    "objects":[
        {
            "class":"body",
            "id":9,
            "object_id":"1299150a-3927-4474-8ffb-fde7b36a5eb1"
        }
    ]
}

Face detection and tracking (face_tracking)

{
    // information about videostream source
    "stream_name":"rtsp://user:password@your/stream/path",
    "stream_title":"your_title",
    "stream_uuid":"3325d094-2e01-4199-931d-876200ed5b68",

    // frame information
    "timestamp_millisec":1720777422166,
    "camera_frame_data": {
        "frame_index":5457,
        "timestamp_millisec":1720777422166
    },
    "image": {
        "color_space":"BGR",
        "dtype":"uint8_t",
        "format":"NDARRAY",
        "shape":[1080,1920,3]
    },
    "is_benchmark_frame":false,

    // information on tracked objects in the frame
    "objects":[
        {
            "__full_frame_bbox":[0.5796875,0.436111,0.615625,0.511111],
            "__index":0,
            "angles":{
                "pitch":-56.2355,
                "roll":4.3686,
                "yaw":36.3418
            },
            "bbox":[0.5796875,0.436111,0.615625,0.511111],
            "class":"face",
            "confidence":1.0,
            "detection_meta": {
                "confirmed":1,
                "good_face_size":1,
                "good_light_and_blur":-1
            },
            "id":16175631120106950574,
            "is_weak":true,
            "quality":-1292.4166,
            "visibility_score":0.9879
        }
    ]
}

Face loss (face_lost)

{
    // information about videostream source
    "stream_name":"rtsp://user:password@your/stream/path",
    "stream_title":"your_title",
    "stream_uuid":"3325d094-2e01-4199-931d-876200ed5b68",

    // frame information
    "timestamp_millisec":1720779397441,
    "camera_frame_data": {
        "frame_index":64664,
        "timestamp_millisec":1720779397441
    },
    "image": {
        "color_space":"BGR",
        "dtype":"uint8_t",
        "format":"NDARRAY",
        "shape":[1080,1920,3]
    },
    "is_benchmark_frame":false,

    // information on lost objects 
    "hashed_face_id":12542825825967542253,
    "objects":[
        {
            "__full_frame_bbox":[0.9140625,0.5453703,0.9401041,0.58518518],
            "angles":{
                "pitch":-12.0049,
                "roll":84.8848,
                "yaw":27.7249
            },
            "bbox":[0.9140625,0.5453703,0.9401041,0.58518518],
            "class":"face",
            "confidence":1.0,
            "id":12542825825967542253,
            "quality":-1480.1163,
            "visibility_score":0.8565
        }
    ]
}

Setting up MQTT sending

1. Set up sending processes via MQTT as described in Processes and Events.
2. In the config/run_params.json file, in the mqtt_settings object, add the enable_task_types array listing the desired raw data types.

The following types are available:

• body_tracking - body detecting and tracking;
• body_lost - losing a previously tracked body;
• face_tracking - face detecting and tracking;
• face_lost - losing a previously tracked face.

Example configuration for sending face_tracking and face_lost:

"mqtt_settings": {
"enable_task_types": [
"face_tracking", "face_lost"
],
// other fields
}