EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

DANet: Joint Density- and Semantics-Adaptive Convolution for 3D Point-Cloud Semantic Segmentation

Weijian Hu iD, Shuning Wang, Lingfang Li iD, Jikai Zhang iD, Ke Han iD

Sensors · 18.07.2026 · 26 (14) · 4561

DOI: 10.3390/s26144561 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 18.07.2026. Verfasst von Weijian Hu und 4 weiteren Autor:innen. Beteiligte Einrichtung: School of Transportation and Logistics, Southwest Jiaotong University, Xian Road, 999, Chengdu 611730, China. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Semantic segmentation of 3D point clouds remains difficult when LiDAR or depth-camera data are sampled unevenly. This paper presents DANet, a 3D semantic segmentation framework built on joint density- and semantics-adaptive convolution. Its core operator, Density-Adaptive Radius Convolution (DAR-Conv), predicts point-wise neighborhood radii before feature aggregation by combining density-driven initialization with semantics-aware modulation. In this way, dense regions can use compact receptive fields, whereas sparse or semantically complex regions can draw on broader contextual support. DANet also includes a Gated Adaptive Cross-Layer Fusion (GACF) module, which aligns encoder–decoder features and performs gated fusion with residual refinement. Experiments on S3DIS and NPM3D show that DANet obtains the highest reported mean accuracy (mAcc) among the compared methods on S3DIS, and high mean Intersection over Union (mIoU) and overall accuracy (OA) on NPM3D, supporting the usefulness of density- and semantics-aware receptive-field adaptation.

Schlagwörter

Stichwörter der Autor:innen

3D perception3D point-cloud semantic segmentationLiDARadaptive convolutiondepth sensingnon-uniform sampling

Autor:innen und Institutionen

  • School of Transportation and Logistics, Southwest Jiaotong University, Xian Road, 999, Chengdu 611730, China.
  • School of Digital and Intelligent Industry, Inner Mongolia University of Science and Technology, 7 Aerding Street, Baotou 014010, China.

ORCID-Profile: Weijian Hu, Lingfang Li, Jikai Zhang, Ke Han

Zitieren

APA

Hu, W., Wang, S., Li, L., Zhang, J., & Han, K. (2026). DANet: Joint Density- and Semantics-Adaptive Convolution for 3D Point-Cloud Semantic Segmentation. Sensors, 26(14), 4561. https://doi.org/10.3390/s26144561

Vancouver

Hu W, Wang S, Li L, Zhang J, Han K. DANet: Joint Density- and Semantics-Adaptive Convolution for 3D Point-Cloud Semantic Segmentation. Sensors. 2026;26(14):4561. doi:10.3390/s26144561

BibTeX
@article{Hu2026,
  author = {Hu, Weijian and Wang, Shuning and Li, Lingfang and Zhang, Jikai and Han, Ke},
  title = {DANet: Joint Density- and Semantics-Adaptive Convolution for 3D Point-Cloud Semantic Segmentation},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {14},
  pages = {4561},
  doi = {10.3390/s26144561}
}
Kostenlos starten →