Enhancing Floor Plan Recognition: A Hybrid Mix-Transformer and U-Net Approach for Precise Wall Segmentation
Dmitriy Parashchuk, Alexey Kapshitskiy, Yuriy Karyakin · 2025
Read Enhancing Floor Plan Recognition: A Hybrid Mix-Transformer and U-Net Approach for Precise Wall Segmentation by Dmitriy Parashchuk, Alexey Kapshitskiy, Yuriy Karyakin (2025) 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-2512-02413.txt.
Description
Turning a flat floor plan into a 3D model starts with finding the walls, thin lines that segmentation models easily miss or blur. Dmitriy Parashchuk, Alexey Kapshitskiy, and Yuriy Karyakin introduce MitUNet in 2025, a hybrid network for precise wall segmentation in floor plan recognition.
The design pairs a Mix-Transformer encoder with a U-Net decoder, reinforced by spatial and channel attention blocks. That combination targets both thin structures and geometric precision, the two known weak points. Accurate automatic reconstruction of indoor spaces feeds renovation planning, real estate tools, and robot navigation maps.
Catalog record
| CREATOR | Dmitriy Parashchuk, Alexey Kapshitskiy, Yuriy Karyakin |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 28 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-02413