Data Fusion Contest (DFC2019)

该数据集包含了由World-View 3卫星拍摄的杰克逊维尔（JAX）地区的三维重建数据，其图像分辨率为0.3米/像素，覆盖了一年的周期。此外，该数据集还使用了机载激光雷达数字表面模型（DSM）作为三维重建的地面真实数据。卫星图像的分辨率为0.3米/像素，而激光雷达DSM的地面采样距离为0.5米。该数据集的任务是利用卫星图像进行三维重建。

This dataset comprises three-dimensional reconstruction data of the Jacksonville (JAX) region captured by the WorldView-3 satellite, with an image resolution of 0.3 meters per pixel and a temporal coverage of one full year. Additionally, airborne LiDAR digital surface model (DSM) is employed as the ground truth for this three-dimensional reconstruction. The satellite imagery has a resolution of 0.3 meters per pixel, while the LiDAR DSM has a ground sampling distance of 0.5 meters. The task of this dataset is to perform three-dimensional reconstruction using satellite imagery.

搜集汇总

背景与挑战

背景概述

DFC2019是一个多赛道的数据融合竞赛数据集，涵盖单视图语义3D、成对语义立体、多视图语义立体和3D点云分类四个任务。数据集提供了丰富的训练和测试数据，包括卫星图像和LiDAR数据，并提供了Python实现的基线算法以帮助参与者快速上手。