SUM Parts: Benchmarking Part-Level Semantic Segmentation of Urban Meshes

Weixiao Gao, Liangliang Nan, Hugo Ledoux
2/10/2026

No official code found. Best implementation: tudelft3d/SUM-Parts-Benchmarks (26 stars, GPL-3.0, last commit 2025-06-21).

Abstract

Semantic segmentation in urban scene analysis has mainly focused on images or point clouds, while textured meshes—offering richer spatial representation—remain underexplored. This paper introduces SUM Parts, the first large-scale dataset for urban textured meshes with part-level semantic labels, covering about 2.5 km2 with 21 classes. The dataset was created using our own annotation tool, which supports both face- and texture-based annotations with efficient interactive selection. We also provide a comprehensive evaluation of 3D semantic segmentation and interactive annotation methods on this dataset. Our project page is available at https://tudelft3d.github.io/SUMParts/.

View on arXivView PDF

Code Implementations(1)

tudelft3d/SUM-Parts-BenchmarksCommunity61%
GPL-3.0

SUM Parts: Benchmarking Part-Level Semantic Segmentation of Urban Meshes (Code and Data)

263C, C++Apr 9, 20251 years agoGPL-3.0

Similar Papers

OpenUrban3D: Annotation-Free Open-Vocabulary Semantic Segmentation of Large-Scale Urban Point Clouds

87% match2025

TrueCity: Real and Simulated Urban Data for Cross-Domain 3D Scene Understanding

85% match2025

FlyAwareV2: A Multimodal Cross-Domain UAV Dataset for Urban Scene Understanding

84% match2025

Turin3D: Evaluating Adaptation Strategies under Label Scarcity in Urban LiDAR Segmentation with Semi-Supervised Techniques

83% match2025

Ultra-High Resolution Segmentation with Ultra-Rich Context: A Novel Benchmark

83% match1 repo352026

ZAHA: Introducing the Level of Facade Generalization and the Large-Scale Point Cloud Facade Semantic Segmentation Benchmark Dataset

83% match1 repo272024

Cite this paper

@article{gao2026parts,
  title  = {SUM Parts: Benchmarking Part-Level Semantic Segmentation of Urban Meshes},
  author = {Weixiao Gao and Liangliang Nan and Hugo Ledoux},
  year   = {2026},
  eprint = {2503.15300},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/2503.15300},
  journal = {CVPR 2025 2025}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2503.15300.svg)](https://codeofpaper.com/paper/2503.15300)

Discussion