The completed task will have a response.point_segmentations.url field, which points to a URL that will contain a JSON containing a list of labels objects, and an array of arrays (one for each frame) point_labels with the point-wise label for each point.
Each element of point_labels is a list of indices of the labels for each point, in the order that they were given to us in the input JSON. Unsegmented points will be given label index 0.
If there are LiDAR annotations associated with the task, the response will also have a response.annotations.url field, which points to a URL that will contain a JSON with the LiDAR annotation data.
{
"annotations": {
"url": "https://scaleapi-results.hosting/com/1"
},
"point_segmentations": {
"url": "https://scaleapi-results.hosting.com/2"
},
}
{
// provided objects and segmentation_labels
"objects": [
{
"label": "Car", // label of the object
"index": 1 // 1-based index into this array of this object (0 is unlabeled)
},
{
"label": "Motorcycle",
"index": 2
},
{
"label": "Cyclist",
"index": 3
},
{
"label": "Adult",
"index": 4
},
{
"label": "Child",
"index": 5
},
{
"label": "Road",
"index": 6
},
{
"label": "Vegetation",
"index": 7
},
{
"label": "Sidewalk",
"index": 8
},
{
"label": "Animal",
"index": 9,
"instance": true
},
{
"label": "Animal",
"index": 10,
"instance": true
}
],
"point_objects": [
[0, 0, 0, 3, 5, 5, 5, 1, 4...], // indexes into the objects array above
[2, 2, 2, 1, 1, 9, 9, 10, 10...] // each array corresponds to a single frame
]
}
Notice
See the Callback Section for more details about callbacks.