ISAT_with_segment_anything: an interactive semi-automatic image segmentation annotation tool based on Segment Anything

What it solves

ISAT is designed to simplify the tedious and time-consuming process of image segmentation annotation. It reduces the manual effort required to create high-quality masks by providing a semi-automatic way to label images.

How it works

The tool leverages the Segment Anything Model (SAM) to allow users to interactively generate segmentation masks. It supports multiple prompting methods to guide the AI in identifying objects:

• Visual Prompts: Users can provide visual cues to guide the segmentation.
• Text Prompts: Users can specify categories (single or multiple) via text to identify objects.
• Plugin System: The software includes a plugin architecture that allows developers to extend its functionality, such as adding auto-annotation capabilities via YOLO models or custom mask export tools.

Who it’s for

It is primarily for researchers and developers working in the field of computer vision and image segmentation who need to create labeled datasets for training machine learning models.

Highlights

• Interactive Annotation: Uses SAM to generate masks semi-automatically.
• Multi-modal Prompting: Supports text and visual prompts for object identification.
• ** uma Plugin Architecture:** Allows for easy extension of functionality with minimal code.
• SAM3 Support: Integrated with the latest SAM3 capabilities.