OmniHorizon
收藏arXiv2023-01-09 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2212.05040v2
下载链接
链接失效反馈官方服务：
资源简介：
OmniHorizon是一个合成数据集，包含24,335个全方向视图，涵盖室内外多种环境，如建筑、街道和多样植被。该数据集考虑了动态场景元素，包括光照变化、不同时间段设置、行人和车辆。创建过程中使用了Unreal Engine 4进行渲染，提供了彩色图像、场景深度和世界法线。数据集主要用于户外场景的深度和法线估计，旨在解决现有数据集在户外场景应用中的局限性，特别是在自动驾驶和导航等领域的应用。
OmniHorizon is a synthetic dataset containing 24,335 omnidirectional views, covering diverse indoor and outdoor environments including buildings, streets, and various vegetation. This dataset incorporates dynamic scene elements such as varying lighting conditions, different time-of-day settings, pedestrians and vehicles. Rendered using Unreal Engine 4 during its creation, the dataset provides color images, scene depth maps, and world normals. It is primarily intended for depth and normal estimation in outdoor scenes, aiming to address the limitations of existing datasets in outdoor scene applications, particularly in fields such as autonomous driving and navigation.
提供机构：
Touchlab, Indian Institute of Technology Madras, Chennai, India
创建时间：
2022-12-10
