segment-anything-model

Segment objects in images using point, box, or mask prompts.

Updated May 13, 2026

Search Tags:#computer vision #object detection #SAM #zero-shot #image segmentation

Ready to run

npx skills add https://github.com/skkeoriw/hermes-brain --skill segment-anything-model-skkeoriw

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: segment-anything-model
Source: https://github.com/skkeoriw/hermes-brain/tree/main/brain/hermes-home/skills/mlops/models/segment-anything
Command: npx skills add https://github.com/skkeoriw/hermes-brain --skill segment-anything-model-skkeoriw

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires segment_anything, opencv-python, pycocotools, matplotlib, onnxruntime, onnx, transformers, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill solves the problem of segmenting any object in images with point, box, or mask prompts without the need for specific task training.

Core Features & Use Cases

Zero-shot Segmentation: Segments objects in any image domain without fine-tuning.
Flexible Prompts: Accepts point, bounding box, or previous mask prompts.
Automatic Segmentation: Generates all object masks automatically.
High Quality: Trained on a large dataset for accuracy.
Model Variants: Offers different model sizes for various use cases.
ONNX Export: Supports deployment in browsers and edge devices.
Use Case: Ideal for object detection, segmentation, medical imaging, and satellite imagery processing.

Quick Start

To segment an object in an image, use the following command:

predictor.predict(point_coords=[x, y], point_labels=[1])

Frequently Asked Questions about segment-anything-model

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I perform zero-shot image segmentation on objects without fine-tuning a model?▼

Zero-shot image segmentation allows you to segment any object in images using point, box, or mask prompts without specific task training. You can generate all object masks automatically by providing coordinates to the predictor.

Can I use point and bounding box prompts to segment specific objects in an image?▼

Yes, you can segment specific objects in an image by providing point coordinates or bounding box prompts. The predictor accepts these inputs to accurately isolate objects without requiring additional fine-tuning.

Does this image segmentation approach support deployment on edge devices and browsers?▼

Yes, the image segmentation approach supports ONNX export for deployment in browsers and edge devices. This allows you to run zero-shot segmentation efficiently using onnxruntime outside of standard Python environments.

What are the best use cases for zero-shot image segmentation with flexible prompts?▼

The best use cases for zero-shot image segmentation include object detection, medical imaging, and satellite imagery processing. It is ideal for scenarios where you need to isolate objects across varying image domains without task-specific training.

How can I automatically generate all object masks in an image without manual prompts?▼

You can automatically generate all object masks in an image by using the automatic segmentation feature. This allows the model to detect and segment every object present without requiring manual point or box inputs.

Do I need to install opencv-python and pycocotools to run zero-shot image segmentation?▼

Yes, you need to install the segment_anything library along with dependencies like opencv-python, pycocotools, and matplotlib to run zero-shot image segmentation. These packages handle image processing and visualization requirements.