[论文] Wat3R: Underwater 3D Geometry Learning without Annotations

论文概要

研究领域: CV
作者: Jiangwei Ren, Xingyu Jiang, Zijie Song
发布时间: 2026-07-10
arXiv: 2507.08184

中文摘要

水下环境中的3D几何估计面临独特挑战：光线衰减、散射，以及缺乏大规模高质量3D标注。现有方法依赖密集标注，在水下场景不切实际。本文提出Wat3R，一种跨域半监督学习框架，将前馈3D重建模型从空气场景迁移到水下场景。独特之处在于，该方法无需任何水下标注数据，采用师生架构，仅凭大量未标注的真实水下视频即可学习鲁棒的几何表征。我们还设计了跨视角一致性损失，利用其他视角的几何线索来补偿当前视角因水衰减和散射造成的信息退化。此外，考虑到缺乏全面的评估基准，我们构建了Water3D数据集，涵盖多种水体和水下场景，用于几何任务评估。实验结果表明，Wat3R在水下多视角深度估计和点云重建方面均优于当前最先进方法。

原文摘要

Estimating 3D geometry in underwater environments presents unique challenges due to light attenuation, scattering, and the absence of large-scale, high-quality 3D annotations. Pioneering methods rely on massive dense annotations that are impractical in underwater settings. In this paper, we propose Wat3R, a cross-domain semi-supervised learning framework designed to adapt feed-forward 3D reconstruction models from air to underwater scenes. Uniquely, our method eliminates the need for any annotated underwater data following a teacher-student architecture, that learns robust geometry representations merely on abundant unlabeled real underwater video footage. We also design a cross-view consistency loss that leverages geometric cues from other视角 to compensate for the information degradation i...

自动采集于 2026-07-11

#论文 #arXiv #CV #小凯