GUIDE NLP Autolabeling with Quality Assurance 🤖

PolygonLabels

The PolygonLabels tag is used to create labeled polygons. Use to apply labels to polygons in semantic segmentation tasks.

Use with the following data types: image.

Parameters

Param Type Default Description
name string Name of tag
toName string Name of image to label
[choice] single | multiple single Configure whether you can select one or multiple labels
[maxUsages] number Maximum number of times a label can be used per task
[showInline] boolean true Show labels in the same visual line
[opacity] number 0.2 Opacity of polygon
[fillColor] string Polygon fill color in hexadecimal
[strokeColor] string Stroke color in hexadecimal
[strokeWidth] number 1 Width of stroke
[pointSize] small | medium | large medium Size of polygon handle points
[pointStyle] rectangle | circle rectangle Style of points
[snap] pixel | none none Snap polygon to image pixels

Sample Results JSON

Name Type Description
original_width number width of the original image (px)
original_height number height of the original image (px)
image_rotation number rotation degree of the image (deg)
value Object
value.points Array.<Array.<number>> list of (x, y) coordinates of the polygon by percentage of the image size (0-100)

Example JSON

{
  "original_width": 1920,
  "original_height": 1280,
  "image_rotation": 0,
  "value": {
    "points": [[2, 2], [3.5, 8.1], [3.5, 12.6]],
    "polygonlabels": ["Car"]
  }
}

Example

Basic labeling configuration for polygonal semantic segmentation of images

<View>
  <Image name="image" value="$image" />
  <PolygonLabels name="labels" toName="image">
    <Label value="Car" />
    <Label value="Sign" />
  </PolygonLabels>
</View>
Designed for teams of all sizes Contact Sales