MultiModal Material Segmentation (MCubeS)
收藏OpenDataLab2026-05-24 更新2024-05-09 收录
下载链接:
https://opendatalab.org.cn/OpenDataLab/MultiModal_Material_Segmentation_MCubeS
下载链接
链接失效反馈官方服务:
资源简介:
MCubeS从道路,人行道或人行道上的角度捕获了日常户外场景中各种材料的视觉外观。在每个视点,我们捕获具有三种根本不同的成像方式的图像,RGB,偏振 (表示为Aolp和Dolp) 和近红外 (NIR)。MCubeS数据集中有500个图像集,RGB图像中的每个像素都被标记为13个材料类之一。
MCubeS captures the visual appearances of diverse materials in everyday outdoor scenes from viewpoints situated on roads, sidewalks or pavements. For each such viewpoint, we acquire images using three fundamentally distinct imaging modalities: RGB, polarimetric (denoted as Aolp and Dolp), and near-infrared (NIR). The MCubeS dataset contains 500 image sets, and every pixel in the RGB images is labeled as one of 13 material classes.
提供机构:
OpenDataLab
创建时间:
2023-02-13
搜集汇总
数据集介绍

以上内容由遇见数据集搜集并总结生成



