SegSpy

📖 Read the documentation

Headless, backend-pluggable EM image segmentation library for HyperSpy.

SegSpy turns a HyperSpy Signal2D into a list of particle objects. It is microscope-generic (TEM/SEM) and particle-type-agnostic, shipping multiple interchangeable segmentation backends behind a uniform registry:

• traditional_cv — CLAHE → black-hat → threshold → morphology → GrabCut refinement
• sam — Meta SAM Auto-Mask Generator with a 5-stage post-filter

Installation

pip install SegSpy

For the SAM backend, install the optional extra:

pip install "SegSpy[sam]"   # adds segment-anything + torch

For development:

git clone https://github.com/VANvonZHANG/SegSpy.git
cd SegSpy
pip install -e ".[dev]"
pre-commit install

Quick start

import hyperspy.api as hs
from SegSpy import SegConfig, SegmentationRegistry, measure_morphology

signal = hs.load("image.dm4")
backend = SegmentationRegistry.get("traditional_cv")
mask = backend.segment(signal, SegConfig())
objects = backend.extract_objects(signal, mask, SegConfig())
for obj in objects:
    measure_morphology(obj, scale_nm=1.0)

CLI

SegSpy run image.dm4 --backend traditional_cv --output ./out

Writes <stem>_particles.csv (per-particle morphology metrics) and <stem>_labeled.png (a labelled mask) to the output directory.

Status

v0.1 — pure library + CLI, no GUI (deferred to a later release). See the CHANGELOG for details. Contributions welcome — see CONTRIBUTING.md.