Skip to main content
This feature is currently in Early Access. Its user experience is still under development and may change as improvements are introduced. Contact support to gain access to SAM 3.
SAM 3 enables you to label one, multiple, or all object instances at the same time, and track it either forwards or backwards across multiple frames.

​
Ontologies

The following table shows the Ontology shapes for which SAM 3 can be used.
Ontology ShapeSAM 3 Availability
Bounding box✅
Rotatable bounding box✅
Polygon✅
Polyline❌
Primitive❌
Keypoint❌
Bitmask✅

​
Modalities

The following table shows the modalities that support SAM 3 tracking.
ModalitiesSAM 3 Availability
Images✅
Videos✅
Image Groups✅
Image Sequences✅
DICOM❌

​
Using SAM 3

SAM 3 is computationally intensive and performance might be slower than other labeling actions.

​
Label All Objects

There are two ways to label all objects of a specific Ontology class. Method 1: Click the Sparkles symbol in the Class.
This action can be undone using CMD / CTRL + Z
Method 2:
  1. Click the Wand icon in the Class.
  2. In the SAM 3 pop-up, ensure the toggle is set to All objects prediction.
  3. Type the name of the class you want to label.
  4. Click Detect all objects.
  1. Optionally, remove unwanted objects using the Delete button on an instance.
  2. Click Done.

​
Label One Or More Objects

  1. Select the Wand icon in the Ontology class you want to label. A SAM 3 pop-up appears.
  2. In the SAM 3 pop-up, ensure the toggle is set to Single-object detection.
  1. Select the objects you want to label by either clicking a single object or clicking and dragging across the area of the frame containing multiple objects.
  2. Optionally track the objects across frames.
  3. Click Save.

​
Detect all with SAM Feature

SAM 3 supports the following prompt modalities:
  • Concept prompts like simple noun phrases and image exemplars.
  • Visual prompts like bitmask, bounding boxes, and points. Encord supports using bounding boxes and points for visual prompts.
The Detect all with SAM feature allows using one labelled object as a visual prompt to SAM 3 to label all similar objects on an image or a video frame using the same class of label. This is useful for classes that are not readily understood by SAM 3 when using a text prompt.
  1. Create a label manually on an object.
  2. Right-click on the label. A menu appears.
  3. Click Detect all with SAM from the menu.
  1. Click Create labels.