From Coarse to Fine: Deep Prototype Refinement Network for Few-Shot Point Cloud Semantic Segmentation

Few-shot point cloud semantic segmentation (FS-PCSS) aims to achieve precise segmentation of novel categories using only limited labeled samples. Existing prototype-based methods rely on shallow feature fusion, failing to adequately model the feature distribution shift between support and query sets, resulting in insufficient prototype adaptation. To address this, we propose the Deep Prototype Refinement Network (DPR-Net), which systematically achieves progressive adaptation by constructing a coarse-to-fine prototype evolution trajectory. Our core Dynamic Prototype Refinement (DPR) module decomposes features into common and distinctive subspaces based on channel activation, enabling targeted adjustment of domain-sensitive features while preserving class-shared semantics. By cascading multiple refinement modules, we construct a prototype trajectory transitioning from support-biased to query-adapted representations, mitigating both under- and over-adaptation. Furthermore, our Mixture of Prototype Experts (MoPE) mechanism treats prototypes at different stages as experts and ensembles their predictions through confidence-driven weighting. Extensive experiments demonstrate state-of-the-art performance with high efficiency. Our code will be available at https://github.com/changshuowang/DPR-Net.

Imagine showing a robot just a few examples of a chair in a 3D room scan, then asking it to find all chairs in a new room. This is the challenge of few-shot 3D segmentation. Current methods try to adapt from examples to new scenes in one step, often failing when the scenes look different. Our method, DPR-Net, works step-by-step instead. At each step, it separates what is common between the example and new scene from what is different, preserving important shape information while adjusting only what needs to change. It then combines predictions from all steps using a smart weighting system. Our approach achieves top results on standard benchmarks while using very few computational resources, bringing us closer to robots that can learn new objects from just a handful of examples.