Class-Prototypes for Contrastive Learning in Weakly-Supervised 3D Point Cloud Segmentation
Rong Li (South China University of Technology),* Anh-Quan CAO (Inria), Raoul de Charette (Inria)
The 33rd British Machine Vision Conference
Abstract
Annotation of large-scale 3D data is notoriously cumbersome and costly. As an alternative, weakly-supervised learning alleviates such a need by reducing the annotation by several order of magnitudes. We propose a novel architecture-agnostic contrastive learning strategy for 3D segmentation. Since contrastive learning requires rich and diverse examples as keys and anchors, we propose a prototype memory bank capturing class-wise global dataset information efficiently into a small number of prototypes acting as keys. An entropy-driven sampling technique then allows us to select good pixels from predictions as anchors. Experiment using a light-weight projection-based backbone shows we outperform baselines on three challenging real-world outdoor datasets, working with as low as 0.001% annotations.Video
Citation
@inproceedings{Li_2022_BMVC,
author = {Rong Li and Anh-Quan CAO and Raoul de Charette},
title = {Class-Prototypes for Contrastive Learning in Weakly-Supervised 3D Point Cloud Segmentation},
booktitle = {33rd British Machine Vision Conference 2022, {BMVC} 2022, London, UK, November 21-24, 2022},
publisher = {{BMVA} Press},
year = {2022},
url = {https://bmvc2022.mpi-inf.mpg.de/0589.pdf}
}
Copyright © 2022
The British Machine Vision Conference is organised by