[论文] Towards Scaling Marine Perception with Synthetic Data

论文概要

研究领域: CV
作者: Haoyu Ma, Onur Bagoren, Anja Sheppard, Elias Fandi, Ashrith Edukulla, Tanner Aslan, Natasha Sieh, Jingyu Song, Katherine A. Skinner
发布时间: 2026-09-17
arXiv: 2609.20680

中文摘要

在具有挑战性的水下环境中，可扩展机器学习受制于带标注真实训练数据的匮乏：采集昂贵费力，大规模真实数据的收集与整理困难重重。仿真数据可以弥合这一鸿沟，支撑水下感知的众多学习型任务。本文扩展了基于 IsaacSim 的水下感知仿真器 OceanSim，为其配备合成数据生成（SDG）流水线，用于训练水下场景模型。该流水线可生成大规模、自动标注、照片级真实的数据集，场景外观、结构与传感器设置均可配置。我们在真实世界海胆检测任务上评估该流水线，研究不同形式的合场景变异对 sim-to-real 性能的影响。基于实验，我们讨论了结果洞见、当前流水线的主要局限，并提出提升水下渲染保真度、场景多样性以及 sim-to-real 泛化评估的未来方向。开源代码见 https://github.com/umfieldrobotics/OceanSim。

原文摘要

Scalable machine learning in challenging underwater environments is strongly limited by the lack of labeled real-world training data. This data is often expensive and laborious to gather, making large-scale real-world data challenging to gather and curate. However, simulated data can help close the gap, enabling many learning-based tasks for underwater perception. In this work, we extend OceanSim, an IsaacSim-based underwater perception simulator, with a Synthetic Data Generation (SDG) pipeline for training models to be used in underwater scenarios. The proposed pipeline enables users to generate large, automatically labeled, photorealistic datasets with configurable scene appearance, structure, and sensor settings. We evaluate the pipeline on a real-world sea urchin detection task and stu...

自动采集于 2026-09-20

#论文 #arXiv #CV #小凯