[Recruitment]
Associate/Assistant Professor in Artificial Intelligence for Space-Enabled Technologies, Durham University

We are looking for applicants in Artificial Intelligence, Computer Vision, Edge Computing, Digital Twins, Human Computer Interaction, User Modelling, Robotics or Resilient Computing with potentials/achievements in informing space applications.

The post hoder will enjoy 1) a permanent (equivalent to US tenured) position at a top 100 university, 2) significantly reduced teaching, 3) a fully-funded PhD, 4) travel budget, 5) chance for a 2-year fully-funded Post-Doc.

U3DS3: Unsupervised 3D Semantic Scene Segmentation

Jiaxu Liu, Zhengdi Yu, Toby P. Breckon and Hubert P. H. Shum
Proceedings of the 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2024

Core A Conference H5-Index: 95# Core A Conference

U3DS3: Unsupervised 3D Semantic Scene Segmentation
‡ According to Core Ranking 2023"

Abstract

Contemporary point cloud segmentation approaches largely rely on richly annotated 3D training data. However, it is both time-consuming and challenging to obtain consistently accurate annotations for such 3D scene data. Moreover, there is still a lack of investigation into fully unsupervised scene segmentation for point clouds, especially for holistic 3D scenes. This paper presents U3DS³, as a step towards completely unsupervised point cloud segmentation for any holistic 3D scenes. To achieve this, U3DS³ leverages a generalized unsupervised segmentation method for both object and background across both indoor and outdoor static 3D point clouds with no requirement for model pre-training, by leveraging only the inherent information of the point cloud to achieve full 3D scene segmentation. The initial step of our proposed approach involves generating superpoints based on the geometric characteristics of each scene. Subsequently, it undergoes a learning process through a spatial clustering-based methodology, followed by iterative training using pseudo-labels generated in accordance with the cluster centroids. Moreover, by leveraging the invariance and equivariance of the volumetric representations, we apply the geometric transformation on voxelized features to provide two sets of descriptors for robust representation learning. Finally, our evaluation provides state-of-the-art results on the ScanNet and SemanticKitti, and competitive result on the S3DIS benchmark datasets.

Downloads

YouTube

Citations

BibTeX

@inproceedings{liu24u3ds3,
 author={Liu, Jiaxu and Yu, Zhengdi and Breckon, Toby P. and Shum, Hubert P. H.},
 booktitle={Proceedings of the 2024 IEEE/CVF Winter Conference on Applications of Computer Vision},
 title={U3DS3: Unsupervised 3D Semantic Scene Segmentation},
 year={2024},
 month={1},
 publisher={IEEE/CVF},
 location={Hawaii, USA},
}

RIS

TY  - CONF
AU  - Liu, Jiaxu
AU  - Yu, Zhengdi
AU  - Breckon, Toby P.
AU  - Shum, Hubert P. H.
T2  - Proceedings of the 2024 IEEE/CVF Winter Conference on Applications of Computer Vision
TI  - U3DS3: Unsupervised 3D Semantic Scene Segmentation
PY  - 2024
Y1  - 1 2024
PB  - IEEE/CVF
ER  - 

Plain Text

Jiaxu Liu, Zhengdi Yu, Toby P. Breckon and Hubert P. H. Shum, "U3DS3: Unsupervised 3D Semantic Scene Segmentation," in Proceedings of the 2024 IEEE/CVF Winter Conference on Applications of Computer Vision, Hawaii, USA, IEEE/CVF, Jan 2024.

Supporting Grants

Similar Research

Li Li, Hubert P. H. Shum and Toby P. Breckon, "Less is More: Reducing Task and Model Complexity for 3D Point Cloud Semantic Segmentation", Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023
Li Li, Khalid N. Ismail, Hubert P. H. Shum and Toby P. Breckon, "DurLAR: A High-fidelity 128-Channel LiDAR Dataset with Panoramic Ambientand Reflectivity Imagery for Multi-Modal Autonomous Driving Applications", Proceedings of the 2021 International Conference on 3D Vision (3DV), 2021

 

 

Last updated on 17 February 2024
RSS Feed