Skip to main content
Some labels are not regions, they are locations. A price tag on a shelf, a defect on a weld, a joint on a body, the corner of a court: one coordinate carries the whole meaning. The Point tool drops a class-colored marker exactly where you click, and nothing is faster for high-volume location labeling.
Available in image projects. Points share the schema with your other annotation types, so a project can mix points, boxes and polygons per class without any special setup.

​
Place points

PixlHub editor with the Price Tag class and Point tool highlighted with step markers on a supermarket scene

Retail example: arm the Price Tag class, pick the Point tool, then click every tag on the shelf.

1

Arm the class

Press the class hotkey or click the class in the left panel. Every point you drop takes this class until you switch.
2

Pick the Point tool

Click the point icon in the floating palette.
3

Click each location

One click, one marker. The class chip renders next to each point and every marker becomes an entry in Object Instances.
Watch the loop on a live task, four price tags in under ten seconds:
Price Tag point markers placed on shelf labels in a PixlHub retail task

Each marker carries its class chip; the instance list keeps them selectable and countable.

​
Adjust and manage

  • Move a point with the Select tool by dragging its marker.
  • Reclassify by selecting the point and pressing another class hotkey.
  • Count coverage from the class panel: the number next to each class is a live count, which is exactly what you want when guidelines say “mark every price tag”.
  • Hide the clutter with the eye toggles in Object Instances when markers start covering the pixels you still need to read.

​
Where points earn their keep

Point accuracy is zoom-dependent. For dense scenes like shelf labels, zoom to 150-200 percent before clicking; a marker that is five pixels off on a tiny tag is a labeling error a reviewer will send back.

​
Frequently asked questions

Yes. Points are independent annotations, so skeleton-style labeling (several named keypoints on one object) works by using one class per keypoint type, or attributes to distinguish them.
As x and y coordinates with class and metadata, alongside your other annotation types. See Exports for format details.
Work class by class: arm one hotkey, sweep the whole image for that class, then switch. Class-switching per object costs more time than a second pass.

​
Next steps

Bounding Box tool

When the extent matters as much as the location.

Image annotation editor

The full editor tour: panels, palette, hotkeys, saving.