Dynamic Prototype Adaptation with Distillation for Few-shot Point Cloud Segmentation
Authors:
- 18 March 2024
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 810-819
Abstract
Few-shot point cloud segmentation seeks to generate per-point masks for previously unseen categories, using only a minimal set of annotated point clouds as reference. Existing prototype-based methods rely on support prototypes to guide the segmentation of query point clouds, but they encounter challenges when significant object variations exist between the support prototypes and query features. In this work, we present dynamic prototype adaptation (DPA), which explicitly learns task-specific prototypes for each query point cloud to tackle the object variation problem. DPA achieves the adaptation through prototype rectification, aligning vanilla prototypes from support with the query feature distribution, and prototype-to-query attention, extracting task-specific context from query point clouds. Furthermore, we introduce a prototype distillation regularization term, enabling knowledge transfer between early-stage prototypes and their deeper counterparts during adaption. By iteratively applying these adaptations, we generate task-specific prototypes for accurate mask predictions on query point clouds. Extensive experiments on two popular benchmarks show that DPA surpasses state-of-the-art methods by a significant margin, e.g., 7.43% and 6.39% under the 2 -way 1 -shot setting on S3DIS and ScanNet, respectively. Code is available at https://github.com/jliu4ai/DPA.Keywords
This publication has 24 references indexed in Scilit:
- Bidirectional Feature Globalization for Few-shot Semantic Segmentation of 3D Point Cloud Scenes2022
- Self-Distillation from the Last Mini-Batch for Consistency Regularization2022
- Few-shot 3D Point Cloud Semantic Segmentation2021
- RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds2020
- PointConv: Deep Convolutional Networks on 3D Point Clouds2019
- Graph Attention Convolution for Point Cloud Semantic Segmentation2019
- 3D Recurrent Neural Networks with Context Fusion for Point Cloud Semantic Segmentation2018
- Large-Scale Point Cloud Semantic Segmentation with Superpoint Graphs2018
- ScanNet: Richly-Annotated 3D Reconstructions of Indoor Scenes2017
- 3D Semantic Parsing of Large-Scale Indoor Spaces2016