SCOPE: Scene-Contextualised Incremental Few-Shot 3D Segmentation

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Findings 2026

Vishal Thengane^{1, 2} Zhaochong An³ Tianjin Huang⁴ Son Lam Phung² Abdesselam Bouzerdoum² Lu Yin¹ Na Zhao⁵ Xiatian Zhu¹

¹University of Surrey, UK ²University of Wollongong, Australia ³University of Copenhagen, Denmark ⁴University of Exeter, UK ⁵Singapore University of Technology and Design, Singapore

vgthengane (at) gmail (dot) com

Paper

ArXiv Code

SCOPE: Scene-Contextualised Incremental Few-Shot 3D Segmentation teaser

Few-shot learning Continual learning Point Cloud Segmentation

Abstract

Incremental Few-Shot (IFS) segmentation aims to learn new categories over time from only a few annotations. Although widely studied in 2D, it remains underexplored for 3D point clouds. Existing methods suffer from catastrophic forgetting or fail to learn discriminative prototypes under sparse supervision, and often overlook a key cue: novel categories frequently appear as unlabelled background in base-training scenes. We introduce SCOPE (Scene-COntextualised Prototype Enrichment), a plug-and-play background-guided prototype enrichment framework that integrates with any prototype-based 3D segmentation method. After base training, a class-agnostic segmentation model extracts high-confidence pseudo-instances from background regions to build a prototype pool. When novel classes arrive with few labelled samples, relevant background prototypes are retrieved and fused with few-shot prototypes to form enriched representations without retraining the backbone or adding parameters. Experiments on ScanNet and S3DIS show that SCOPE achieves SOTA performance, improving novel-class IoU by up to 6.98% and 3.61%, and mean IoU by 2.25% and 1.70%, respectively, while maintaining low forgetting.

Method

Briefly explain the method, model design, and training strategy.

Results

Summarize key quantitative/qualitative results and important takeaways.

BibTeX

@inproceedings{thengane2026scope,
  title={SCOPE: Scene-Contextualized Incremental Few-Shot 3D Segmentation},
  author={Thengane, Vishal and An, Zhaochong and Huang, Tianjin and Phung, Son Lam and Bouzerdoum,    Abdesselam and Yin, Lu and Zhao, Na and Zhu, Xiatian},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Findings},
  year={2026}
}