| Property |
Type |
Description |
| column.id |
String |
Column identifier |
| column.parent |
String | null |
Parent identifier |
| column.target |
"tasks" | "annotations" |
Entity the column is attached to |
| column.title |
String |
Human readable title |
| column.type |
ColumnType |
Column value type |
| column.children |
List | null
| Column identifier |
| column.visibility_defaults |
Dict<"explore"|"labeling", Boolean> |
Column identifier |
ColumnType
Represents a type of a column value. Column can have one of the tipes listed below.
| Type | Description |
| --- | - |
| String | Primitive string |
| Boolean | Primitive boolean |
| Number | Int or Double |
| Datetime | Date and time in ISO format |
| List | List of items. T can be one of the types listed in this table |
| Image | Image url |
| Audio | Audio url |
| Text | Text string |
| HyperText | HTML or XML based markup |
| TimeSeries | TimeSeries data |
| Unknown | Type cannot be determined by the backend |
ColumnAlias
ColumnAlias is an aggregated field that combines full path to the column. ColumnAlias is built using the following rules:
[target]:[column_name]
[target]:[full_path].[column_name]
Full path is a path that combines all parent columns. E.g. you have a data column that contains several values:
{
"data": {
"image": "https://example.com/image.jpg"
}
}
In this case columns will look like this:
{
"columns": [
{
"id": "data",
"title": "Data",
"children": ["image"],
"target": "tasks",
},
{
"id": "image",
"title": "image",
"parent": "data",
"target": "tasks",
}
]
}
As the columns list is flat, full path will reference all ascending columns: tasks:data.image
In some cases ColumnAlias might be negative. For example negative values are used for ordering: -tasks:data.image
SelectedItems
| Property |
Type |
Default |
Description |
| all |
Boolean |
false |
When true, all items in the dataset implied as selected |
| included |
List
| [] |
When all=false this list specifies selected items |
| excluded |
List
| [] |
When all=true this list specifies the items to exclude from selection |
Selected items is an object that stores samples checked in the UI. To operate effectively on large amounts of data it uses partial selection approach. The structure of this object is the following:
// In this case we select only items with IDs 1, 2 and 3
{
"selectedItems": {
"all": false,
"included": [1, 2, 3]
}
}
// SelectedItems can select all the items in the dataset regardless of the size
{
"selectedItems": {
"all": true,
}
}
// With `excluded` list you can select all but `excluded`
// In this example we select all items except 1, 2 and 3
{
"selectedItems": {
"all": true,
"excluded": [1, 2, 3]
}
}
Task
Task is a single sample from the dataset.
| Property |
Type |
Description |
| id |
Int |
Task identifier generated by LS |
| cancelled_annotations |
Int |
Number of cancelled (rejected) annotations for the task |
| completed_at |
DateTime |
Creation date of the last annotation |
| predictions_result |
String |
|
| total_annotations |
Int |
Total annotations for the task |
| total_predictions |
Int |
Total predictions fo the task |
| data |
Dict
| Data from the dataset |
| extra |
Dict
| Any extra data for the task |