Decoupling 3-D Point Cloud Attributes for Semantic Segmentation via Real-World Prior Exploitation
- 11 March 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 27 (4) , 4004-4016
Abstract
Point cloud semantic segmentation, which involves assigning a category for each point, is a crucial task in autonomous driving and intelligent transportation systems. Due to the inherently unordered and irregular nature of point clouds, learning robust features that accurately capture real-world distributions from point coordinates and other attributes remains challenging. Following the pioneering work of PointNet, current 3D deep neural networks process point coordinates alongside other attributes without fully exploiting the implicit class prior information embedded in spatial information. In this work, we first conduct a pilot study to evaluate how current 3D networks utilize point coordinates and validate the presence of implicit class priors within them. Subsequently, we design a robust Position-to-Physics (P2P) fusion strategy that learns adaptive weights to dynamically incorporate implicit class priors present in point coordinates into point features. Moreover, we design a dual-branch network architecture and propose a triplet loss to further enhance the adaptive fusion process. Extensive experiments demonstrate that decoupling position attributes from physics attributes facilitates the extraction and utilization of implicit class priors. Our proposed modules consistently improve segmentation performance across various networks and datasets, demonstrating their generalizability and effectiveness.Keywords
Funding Information
- National Natural Science Foundation of China (U23A20343)
- National Natural Science Foundation of China (62222602)
- National Natural Science Foundation of China (62206293)
- National Natural Science Foundation of China (62302167)
- National Natural Science Foundation of China (62176224)
- National Natural Science Foundation of China (62006139)
- Shanghai Sailing Program (23YF1410500)
- Natural Science Foundation of Shanghai (23ZR1420400)
- Science and Technology Commission of Shanghai (21511100700)
This publication has 48 references indexed in Scilit:
- P2P: Part-to-Part Motion Cues Guide a Strong Tracking Framework for LiDAR Point CloudsInternational Journal of Computer Vision, 2025
- A Closer Look at Local Aggregation Operators in Point Cloud Analysis2020
- Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution2020
- Semi-supervised Semantic Segmentation via Strong-Weak Dual-Branch Network2020
- 3DFeat-Net: Weakly Supervised Local 3D Features for Point Cloud Registration2018
- PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation2017
- Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs2017
- ScanNet: Richly-Annotated 3D Reconstructions of Indoor Scenes2017
- 3D Semantic Parsing of Large-Scale Indoor Spaces2016
- 3D ShapeNets: A deep representation for volumetric shapes2015