Getting Started

Micro-Seg is a native desktop tool for annotating and analyzing 2D microscopy images, with AI-assisted segmentation that runs entirely on your machine.

System requirements

Windows 10 or later (x64), or Linux (x64)
A GPU is not required, but the interface is GPU-accelerated and benefits from a working graphics driver

Install

Micro-Seg ships as a portable archive — there is no installer.

Get the archive for your platform from the download page: .zip for Windows, .tar.gz for Linux.
Verify the checksum against the checksum file provided with the download (optional but recommended).
Extract the archive anywhere and run the microseg executable.

The ONNX Runtime libraries used for AI segmentation are bundled inside the archive. Model weights are managed separately inside the app.

Create your first project

Create a project. A project is a plain directory on disk — no database, no hidden state elsewhere.
Import images (Ctrl+O). Every imported image is content-addressed by the SHA-256 hash of its raw bytes, so re-importing the same file is detected and rejected as a duplicate.
Define labels. Set up the label categories (name + color) you want to annotate with — they’re on the Labels tab of the right panel.
Annotate. Pick a tool from the left toolbar — see Annotation tools — or let AI segmentation propose masks you refine by hand.
Save (Ctrl+S). Annotations are written crash-safely next to the images inside the project directory. Auto-save is available in the configuration.

Find your way around

The user interface tour explains every region of the main window.
Press F1 anytime for the built-in shortcut list, or read the shortcuts reference.
Annotations export as JSON, COCO or YOLO; the on-disk project format is documented and stable, so pipelines can also read it directly.

Everything the project needs lives inside its directory, which makes projects trivial to copy, archive, or version with Git.