WoodScape：面向自动驾驶的多任务多摄像头鱼眼数据集
计算机视觉与模式识别
2021-07-06 v3 人工智能
机器学习
机器人学
机器学习
摘要
鱼眼相机因能获得大视场角，常用于监控、增强现实尤其是汽车应用中。尽管其应用广泛，但用于细致评估鱼眼图像上计算机视觉算法的公开数据集很少。我们发布了首个大规模的鱼眼汽车数据集 WoodScape，其名称源于1906年发明鱼眼相机的 Robert Wood。WoodScape 包含四个环视摄像头和九项任务，包括分割、深度估计、3D 边界框检测以及污渍检测。针对超过10,000张图像提供了40个类别的实例级语义标注，其他任务的标注则覆盖超过100,000张图像。通过 WoodScape，我们希望鼓励学界为鱼眼相机适配计算机视觉模型，而非使用朴素的校正方法。
引用
@article{arxiv.1905.01489,
title = {WoodScape: A multi-task, multi-camera fisheye dataset for autonomous driving},
author = {Senthil Yogamani and Ciaran Hughes and Jonathan Horgan and Ganesh Sistu and Padraig Varley and Derek O'Dea and Michal Uricar and Stefan Milz and Martin Simon and Karl Amende and Christian Witt and Hazem Rashed and Sumanth Chennupati and Sanjaya Nayak and Saquib Mansoor and Xavier Perroton and Patrick Perez},
journal= {arXiv preprint arXiv:1905.01489},
year = {2021}
}
备注
Accepted for Oral Presentation at IEEE International Conference on Computer Vision (ICCV) 2019. Please refer to our website https://woodscape.valeo.com and https://github.com/valeoai/woodscape for release status and updates