PointSFDA: Source-free Domain Adaptation for Point Cloud Completion

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: He, Xing, Zhu, Zhe, Nan, Liangliang, Chen, Honghua, Qin, Jing, Wei, Mingqiang
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915204879613952
author He, Xing
Zhu, Zhe
Nan, Liangliang
Chen, Honghua
Qin, Jing
Wei, Mingqiang
author_facet He, Xing
Zhu, Zhe
Nan, Liangliang
Chen, Honghua
Qin, Jing
Wei, Mingqiang
contents Conventional methods for point cloud completion, typically trained on synthetic datasets, face significant challenges when applied to out-of-distribution real-world scans. In this paper, we propose an effective yet simple source-free domain adaptation framework for point cloud completion, termed \textbf{PointSFDA}. Unlike unsupervised domain adaptation that reduces the domain gap by directly leveraging labeled source data, PointSFDA uses only a pretrained source model and unlabeled target data for adaptation, avoiding the need for inaccessible source data in practical scenarios. Being the first source-free domain adaptation architecture for point cloud completion, our method offers two core contributions. First, we introduce a coarse-to-fine distillation solution to explicitly transfer the global geometry knowledge learned from the source dataset. Second, as noise may be introduced due to domain gaps, we propose a self-supervised partial-mask consistency training strategy to learn local geometry information in the target domain. Extensive experiments have validated that our method significantly improves the performance of state-of-the-art networks in cross-domain shape completion. Our code is available at \emph{\textcolor{magenta}{https://github.com/Starak-x/PointSFDA}}.
format Preprint
id arxiv_https___arxiv_org_abs_2503_15144
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PointSFDA: Source-free Domain Adaptation for Point Cloud Completion
He, Xing
Zhu, Zhe
Nan, Liangliang
Chen, Honghua
Qin, Jing
Wei, Mingqiang
Computer Vision and Pattern Recognition
Conventional methods for point cloud completion, typically trained on synthetic datasets, face significant challenges when applied to out-of-distribution real-world scans. In this paper, we propose an effective yet simple source-free domain adaptation framework for point cloud completion, termed \textbf{PointSFDA}. Unlike unsupervised domain adaptation that reduces the domain gap by directly leveraging labeled source data, PointSFDA uses only a pretrained source model and unlabeled target data for adaptation, avoiding the need for inaccessible source data in practical scenarios. Being the first source-free domain adaptation architecture for point cloud completion, our method offers two core contributions. First, we introduce a coarse-to-fine distillation solution to explicitly transfer the global geometry knowledge learned from the source dataset. Second, as noise may be introduced due to domain gaps, we propose a self-supervised partial-mask consistency training strategy to learn local geometry information in the target domain. Extensive experiments have validated that our method significantly improves the performance of state-of-the-art networks in cross-domain shape completion. Our code is available at \emph{\textcolor{magenta}{https://github.com/Starak-x/PointSFDA}}.
title PointSFDA: Source-free Domain Adaptation for Point Cloud Completion
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2503.15144