Metadata-Version: 2.4
Name: fastncut
Version: 2.0.5
Summary: An agnostic algorithm for bipartitioning of feature maps. Really fast, even for high resolutions.
Project-URL: Documentation, https://www.agentspace.org/fastncut/fastncut.html
Project-URL: Source, https://github.com/andylucny/fastncut
Project-URL: Tracker, https://github.com/andylucny/fastncut/issues
Requires-Python: >=3.8
Description-Content-Type: text/markdown

# fastncut

A Fast Algorithm for Normalized Cut with Applications on Bipartitioning Feature Maps in Deep Learning

A fast agnostic algorithm for bipartitioning images or feature maps.<BR/>Implemented in Pytorch.<BR/>Really cool!<BR/>The ingenious idea of Shi & Malik was brought into practice.
See <a href="https://github.com/andylucny/fastncut">https://github.com/andylucny/fastncut</a>

26.3.2026, 2.0.4, initial release on PyPY with documentation<BR/>
2.4.2026, 2.0.5, support of data formats nc and bnc<BR/>
