Saved in:
Bibliographic Details
Main Authors: Laudari, Sudip, Baek, Sang Hun
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.14764
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916015174057984
author Laudari, Sudip
Baek, Sang Hun
author_facet Laudari, Sudip
Baek, Sang Hun
contents Geometric data augmentation is widely used in segmentation workflows, but polygon annotations are often assumed to remain valid after transformation. This assumption can fail in structured domains such as architectural floorplan analysis, where a region may contain an interior void encoded as part of a single ordered polygon chain. Cropping or clipping can remove bridge vertices in this chain, causing one semantic region to split into disconnected components. We propose a lightweight topology-preserving augmentation strategy that repairs missing adjacency relations in index space while preserving the original vertex order. The method adds minimal overhead and can be integrated into existing preprocessing workflows. Experiments show that the proposed approach achieves near-perfect Cyclic Adjacency Preservation (CAP) across common geometric transformations and improves annotation consistency in polygon-based segmentation.
format Preprint
id arxiv_https___arxiv_org_abs_2603_14764
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Topology-Preserving Polygon Augmentation for Segmentation in Structured Visual Domains
Laudari, Sudip
Baek, Sang Hun
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Geometric data augmentation is widely used in segmentation workflows, but polygon annotations are often assumed to remain valid after transformation. This assumption can fail in structured domains such as architectural floorplan analysis, where a region may contain an interior void encoded as part of a single ordered polygon chain. Cropping or clipping can remove bridge vertices in this chain, causing one semantic region to split into disconnected components. We propose a lightweight topology-preserving augmentation strategy that repairs missing adjacency relations in index space while preserving the original vertex order. The method adds minimal overhead and can be integrated into existing preprocessing workflows. Experiments show that the proposed approach achieves near-perfect Cyclic Adjacency Preservation (CAP) across common geometric transformations and improves annotation consistency in polygon-based segmentation.
title Topology-Preserving Polygon Augmentation for Segmentation in Structured Visual Domains
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2603.14764