Once the lidarlinking
task is done, we will send a response to your callback URL. The response.annotations.url
field of the callback will be a link to a JSON file containing the annotations.
{
"response": {
"annotations": {
"url": "https://s3-us-west-1.amazonaws.com/scaleapi-results/000003f0-a07c-11e8-891a-b1d8c6329c00"
}
},
"task_id": "5967d6b572633603819b188d",
"task": { ... }
}
The annotation format is similar to the callback format of videoannotation
tasks: it will be a list of LidarLinkingFrame objects, one for each frame of the lidar task. The LidarLinkingFrame objects will be in temporal order; that is, the first LidarLinkingFrame will correspond to the first frame in the lidarannotation
task's attachments array, etc. Each LidarLinkingFrame has a camera_responses
field referencing a list of CameraResponse objects. Each CameraResponse object has a annotations
field, containing a reference to an AnnotationFrame, and a camera_used
field, which specifies the camera number that the annotations in the CameraResponse appear in.
Each AnnotationFrame is a list of Annotation objects, which has the same structure as the Annotation objects in videoannotation
tasks except for two exceptions. cuboid_uuid
is the cuboid ID for the corresponding cuboid in the paired lidarannotation
task (or is a random uuid for newly drawn annotations) - this is used instead of the uuid
field. not_in_lidar
is set to true for annotations that do not appear in the paired lidarannotation
task.
[
{ // annotations in frame 0
"camera_responses": [
{ // annotations for frame 0, camera 1
"annotations": [
{
"cuboid_uuid": "0f81f763-a35c-4b11-9791-019f8cda6524",
"label": "car",
"type": "box",
// geometry-specific annotation fields
"top": 200,
"left": 105,
"width": 280,
"height": 192
},
...
],
"camera_used": 1,
},
{ // annotations for frame 0, camera 2
"annotations": [
{
"cuboid_uuid": "0f81f763-a35c-4b11-9791-019f8cda6524",
"label": "car",
"type": "box",
// geometry-specific annotation fields
"top": 250,
"left": 155,
"width": 280,
"height": 192
},
...
],
"camera_used": 2,
},
...
]
},
{ // annotations in frame 1
"camera_responses": [
{ // annotations for frame 1, camera 1
"annotations": [
{
"cuboid_uuid": "0f81f763-a35c-4b11-9791-019f8cda6524",
"label": "car",
"type": "box"
// geometry-specific annotation fields
"top": 100,
"left": 205,
"width": 280,
"height": 192
},
...
],
"camera_used": 1,
},
...
]
},
]
Notice
See the Callback Section for more details about callbacks.