Learning with Geometric Priors in U-Net Variants for Polyp Segmentation
Fabian Vazquez, Jose A. Nuñez, Diego Adame, Alissen Moreno, Augustin Zhan, Huimin Li, Jinghao Yang, Haoteng Tang, Bin Fu, Pengfei Gu · 2026
Read Learning with Geometric Priors in U-Net Variants for Polyp Segmentation by Fabian Vazquez, Jose A. Nuñez, Diego Adame, Alissen Moreno, Augustin Zhan, Huimin Li, Jinghao Yang, Haoteng Tang, Bin Fu, Pengfei Gu (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-2601-17331.txt.
Description
Fabian Vazquez and seven co-authors work on finding polyps in colonoscopy images. Polyps can grow into colorectal cancer, so early finds save lives. Computer tools help doctors spot them in scans. U-Net style models, in CNN, Transformer, and Mamba forms, already score well. They still miss shape and structure cues. Low contrast and clutter make those cues hard to read.
The team adds geometric priors to guide the models. Priors tell the network what shapes to expect. That help should sharpen borders in messy scenes. Sharper maps mean fewer missed polyps. The work serves computer-aided diagnosis in real clinics. It aims at the cases where current tools are weakest.
Catalog record
| CREATOR | Fabian Vazquez, Jose A. Nuñez, Diego Adame, Alissen Moreno, Augustin Zhan, Huimin Li, Jinghao Yang, Haoteng Tang, Bin Fu, Pengfei Gu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 19 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-17331