UAV-LADIS dataset

A custom dataset\u2014UAV-LADIS\u2014comprising both an aerial image dataset and a 3D occlusion scenario dataset. The image dataset was captured using a DJI Mini 3 drone over urban and suburban areas in Shanghai, China. It consists of 10 video sequences at a resolution of 1920\u00d71080, recorded at a flight altitude of 90 meters. The captured scenes feature large-scale low-texture regions (e.g., rooftops, asphalt roads) and complex occlusions caused by roadside vegetation and urban infrastructures. The 3D occlusion scenario dataset primarily targets occlusions induced by roadside-trees and bridges, encompassing a total of 212 synthetic occluded scenes. Specifically, 57 bridge models were collected to construct occlusion scenarios involving bridges. Additionally, 31 roadside-tree models were acquired and procedurally combined using SketchUp (SU) software to synthesize 155 tree-based occlusion scenes. These occlusion scenarios were rendered and voxelized using the open-source software Blender to generate top-down depth maps and their corresponding 3D voxel grids. The UAV-LADIS dataset is publicly available at: \\url{https:\/\/github.com\/Qianwei-SHU\/UAV-LADIS}.