ComfyUI-segment-anything-2

DownloadAndLoadSAM2Model

Loads a selected SAM2 model variant with the chosen segmentor, device, and precision, downloading the model into ComfyUI's sam2 models directory when it is not present.

SAM2 · 0 inputs · 4 parameters · 1 output

Florence2toCoordinates

Selects bounding boxes from JSON data and returns their center coordinates as a JSON string along with the selected bounding boxes; batch mode can process the selected indices across the data items.

SAM2 · 1 input · 2 parameters · 2 outputs

Sam2AutoSegmentation

Runs SAM2 automatic mask generation on an image batch using the configured point sampling, prediction and stability thresholds, crop settings, and mask options; returns masks, a colored segmented image, and bounding boxes.

SAM2 · 2 inputs · 14 parameters · 3 outputs

Sam2Segmentation

Accepts a SAM2 model and image with optional positive or negative point coordinates, bounding boxes, and a mask, and returns a mask.

SAM2 · 6 inputs · 2 parameters · 1 output

Sam2VideoSegmentation

Propagates SAM2 video segmentation through the frames in an inference state, combines object masks per frame, and returns a mask batch.

SAM2 · 2 inputs · 1 parameter · 1 output

Sam2VideoSegmentationAddPoints

Adds positive and optional negative point prompts for a selected frame and object to a SAM2 video inference state, initializing or reusing state, and returns the model and inference state.

SAM2 · 5 inputs · 2 parameters · 2 outputs