graph-segment 0.0.1

Graph-based image segmentation

Graph Segment Algorithms for Graph Neural Network-based Tracking

Installation

pip install .

Test

Download the data python3 download_data.py. And then ./build/bin/walk_through data/debug_graph.dot.

Developer Guide

The following commands are for Perlmutter. For other systems, please adjust the commands accordingly.

podman-hpc pull docker.io/docexoty/mltools:20250227

Or you can build the image from the Dockerfile in this repository.

Then run the following command to start the container, and build the code:

podman-hpc run -it --rm --gpu -v $PWD:$PWD -w $PWD docexoty/mltools:20250227 bash

cmake -B build -S . -Dpybind11_DIR=/usr/local/lib/python3.10/dist-packages/pybind11/share/cmake/pybind11 -DCMAKE_BUILD_TYPE=DEBUG
cmake --build build

Packing and Uploading

python3 -m build --sdist
python3 -m build --wheel
twine upload dist/*

Note

The code was improved by OpenAI o1Pro model. The execution time was reduced from 403 ms to 190 ms.