SAM2 Segmentation¶
SAM2 Segment¶
Interactive SAM2 segmentation — click to include/exclude regions.
Details
Connect an image, then click on objects to segment them. Use "+ Object" to add multiple objects, each with a distinct label. Toggle between Include (+foreground) and Exclude (−background) modes with the toolbar button.
Outputs: binary mask (union of all objects), integer label image (each object = unique label), and a colored overlay.
| Direction | Port | Type |
|---|---|---|
| Input | image |
image |
| Output | mask |
mask |
| Output | label_image |
label_image |
| Output | overlay |
overlay |
Properties: Model
Grounding DINO¶
Detect objects by text description using GroundingDINO.
Details
Type a text query (e.g. "nucleus" or "cell, membrane") and get bounding-box detections as rectangular masks.
| Direction | Port | Type |
|---|---|---|
| Input | image |
image |
| Output | mask |
mask |
| Output | label_image |
label_image |
| Output | overlay |
overlay |
Properties: Min Score
SAM2 Text Segment¶
Text-prompted segmentation using GroundingDINO + SAM2.
Details
Type a text description to detect and segment objects. Chains GroundingDINO (text → boxes) with SAM2 (boxes → precise masks).
| Direction | Port | Type |
|---|---|---|
| Input | image |
image |
| Output | mask |
mask |
| Output | label_image |
label_image |
| Output | overlay |
overlay |
Properties: Min Score, SAM2 Model