SCOPE: Scene-Contextualized Incremental Few-Shot 3D Segmentation
This addresses the problem of learning new categories from few annotations in 3D point clouds for applications like robotics and autonomous systems, offering a plug-and-play solution that reduces forgetting and improves accuracy.
The paper tackles incremental few-shot 3D segmentation by proposing SCOPE, a framework that enriches prototypes using background pseudo-instances, achieving state-of-the-art results with improvements in novel-class IoU of up to 6.98% and mean IoU of up to 2.25% on datasets like ScanNet and S3DIS.
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. Code is available https://github.com/Surrey-UP-Lab/SCOPE.