Semantic Segmentation for Histopathology using Learned Regularization based on Global Proportions
Yangping Li ID, Thomas Pinetz ID, Michael Hölzel ID, Marieta Toma, Alexander Effland ID · 2026
Read Semantic Segmentation for Histopathology using Learned Regularization based on Global Proportions by Yangping Li ID, Thomas Pinetz ID, Michael Hölzel ID, Marieta Toma, Alexander Effland ID (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-24347.txt.
Description
Li and four colleagues infer tissue segmentations from label proportions, without pixel labels. Many segmentations can satisfy the same proportions, so the task is unclear. Their Variational Segmentation from Label Proportions method runs a ready net with test-time boosts for scores. A second stage fuses scores through energy tuning with an earth-mover fidelity term and a smoother. Unlike end-to-end networks, it shows its fidelity-control energy. It beats weakly taught and untaught methods on two public datasets.
Catalog record
| CREATOR | Yangping Li ID, Thomas Pinetz ID, Michael Hölzel ID, Marieta Toma, Alexander Effland ID |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49657 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.24347