Segment Anything 2 Model

Class: SegmentAnything2BlockV1

Source: inference.core.workflows.core_steps.models.foundation.segment_anything2.v1.SegmentAnything2BlockV1

Run Segment Anything 2, a zero-shot instance segmentation model, on an image.

** Dedicated inference server required (GPU recomended) **

You can use pass in boxes/predictions from other models to Segment Anything 2 to use as prompts for the model. If you pass in box detections from another model, the class names of the boxes will be forwarded to the predicted masks. If using the model unprompted, the model will assign integers as class names / ids.

Type identifier

Use the following identifier in step "type" field: roboflow_core/segment_anything@v1to add the block as as step in your workflow.

Properties

Name	Type	Description	Refs
name	str	Enter a unique identifier for this step..	❌
version	str	Model to be used. One of hiera_large, hiera_small, hiera_tiny, hiera_b_plus.	✅
threshold	float	Threshold for predicted masks scores.	✅
multimask_output	bool	Flag to determine whether to use sam2 internal multimask or single mask mode. For ambiguous prompts setting to True is recomended..	✅

The Refs column marks possibility to parametrise the property with dynamic values available in workflow runtime. See Bindings for more info.

Available Connections

Compatible Blocks

Check what blocks you can connect to Segment Anything 2 Model in version v1.

• inputs: Trace Visualization, Detections Consensus, Background Color Visualization, VLM As Classifier, Florence-2 Model, Reference Path Visualization, Corner Visualization, Bounding Rectangle, SAM 3, Seg Preview, Perspective Correction, Detection Offset, Roboflow Dataset Upload, OpenAI, Model Monitoring Inference Aggregator, Image Threshold, Pixelate Visualization, Object Detection Model, Email Notification, Cosine Similarity, Label Visualization, Image Slicer, Object Detection Model, Bounding Box Visualization, Google Vision OCR, Crop Visualization, VLM As Detector, Time in Zone, Image Blur, Path Deviation, Anthropic Claude, SIFT, Triangle Visualization, Detection Event Log, Instance Segmentation Model, Gaze Detection, Stitch Images, SIFT Comparison, Heatmap Visualization, SAM 3, Time in Zone, Mask Visualization, Detections Stitch, Email Notification, Stability AI Outpainting, Time in Zone, LMM For Classification, Segment Anything 2 Model, Google Gemini, Velocity, Anthropic Claude, Google Gemini, Dynamic Zone, Template Matching, Google Gemini, Keypoint Detection Model, Twilio SMS Notification, Byte Tracker, Roboflow Dataset Upload, Stability AI Image Generation, SIFT Comparison, Polygon Zone Visualization, YOLO-World Model, Depth Estimation, Dot Visualization, Llama 3.2 Vision, PTZ Tracking (ONVIF).md), Dynamic Crop, Detections Merge, Contrast Equalization, Circle Visualization, Slack Notification, Color Visualization, Image Slicer, Stability AI Inpainting, OpenAI, Clip Comparison, Detections Transformation, Byte Tracker, Local File Sink, JSON Parser, Relative Static Crop, Detections Filter, Instance Segmentation Model, Polygon Visualization, Ellipse Visualization, OCR Model, CogVLM, VLM As Detector, SAM 3, Stitch OCR Detections, Twilio SMS/MMS Notification, Stitch OCR Detections, Moondream2, Halo Visualization, Icon Visualization, Anthropic Claude, Image Contours, Morphological Transformation, Motion Detection, Line Counter, Blur Visualization, Detections List Roll-Up, Detections Combine, OpenAI, Polygon Visualization, CSV Formatter, Detections Stabilizer, Camera Calibration, Detections Classes Replacement, Single-Label Classification Model, Line Counter Visualization, Camera Focus, OpenAI, Webhook Sink, Image Convert Grayscale, LMM, Multi-Label Classification Model, Camera Focus, Classification Label Visualization, Model Comparison Visualization, EasyOCR, Image Preprocessing, VLM As Classifier, Grid Visualization, Background Subtraction, Halo Visualization, Overlap Filter, Keypoint Visualization, QR Code Generator, Florence-2 Model, Identify Outliers, Text Display, Identify Changes, Roboflow Custom Metadata, Byte Tracker, Mask Area Measurement, Keypoint Detection Model, Path Deviation, Absolute Static Crop
• outputs: Trace Visualization, Detections Consensus, Mask Visualization, Detections Stabilizer, Circle Visualization, Background Color Visualization, Size Measurement, Detections Stitch, Color Visualization, Florence-2 Model, Time in Zone, Detections Classes Replacement, Segment Anything 2 Model, Stability AI Inpainting, Bounding Rectangle, Corner Visualization, Velocity, Byte Tracker, Detections Transformation, Dynamic Zone, Detection Offset, Roboflow Dataset Upload, Perspective Correction, Detections Filter, Model Monitoring Inference Aggregator, Pixelate Visualization, Polygon Visualization, Ellipse Visualization, Label Visualization, Byte Tracker, Camera Focus, Model Comparison Visualization, Bounding Box Visualization, Polygon Visualization, Crop Visualization, Roboflow Dataset Upload, Time in Zone, Distance Measurement, Detections Merge, Halo Visualization, Path Deviation, Overlap Filter, Line Counter, Halo Visualization, Florence-2 Model, Icon Visualization, Triangle Visualization, Line Counter, Detection Event Log, Dot Visualization, Heatmap Visualization, Roboflow Custom Metadata, Blur Visualization, Detections List Roll-Up, Byte Tracker, Dynamic Crop, Detections Combine, PTZ Tracking (ONVIF).md), Time in Zone, Mask Area Measurement, Path Deviation

Input and Output Bindings

The available connections depend on its binding kinds. Check what binding kinds Segment Anything 2 Model in version v1 has.

Bindings

• input

  • images (image): The image to infer on..
  • boxes (Union[keypoint_detection_prediction, object_detection_prediction, instance_segmentation_prediction]): Bounding boxes (from another model) to convert to polygons.
  • version (string): Model to be used. One of hiera_large, hiera_small, hiera_tiny, hiera_b_plus.
  • threshold (float): Threshold for predicted masks scores.
  • multimask_output (boolean): Flag to determine whether to use sam2 internal multimask or single mask mode. For ambiguous prompts setting to True is recomended..

• output

  • predictions (instance_segmentation_prediction): Prediction with detected bounding boxes and segmentation masks in form of sv.Detections(...) object.

Example JSON definition of step Segment Anything 2 Model in version v1

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/segment_anything@v1",
    "images": "$inputs.image",
    "boxes": "$steps.object_detection_model.predictions",
    "version": "hiera_large",
    "threshold": 0.3,
    "multimask_output": true
}