Skip to main content

Project Overview

This tool is built on Meta’s Segment Anything Model 2 (SAM2) and provides an interactive image segmentation annotation interface. Running in a Jupyter Notebook environment, researchers can leverage SAM2’s zero-shot segmentation capability to quickly and precisely label target regions in images, making it well-suited for computer vision tasks that require large annotated datasets.

Tech Stack

  • Environment: Jupyter Notebook
  • Core model: Meta SAM2 (Segment Anything Model 2)
  • Language: Python
  • Domain: Image segmentation, data annotation, computer vision dataset construction

Features

  • Integrates SAM2 for zero-shot interactive segmentation
  • Notebook-based workflow for step-by-step operation and parameter adjustment
  • Rapid mask generation for multiple target regions within a single image
  • Annotation results exportable for downstream training pipelines
  • Applicable to semantic labeling of custom datasets

Source Code

GitHub Repository

felimet/SAM2_Annotation_Tool