Voxel51/widedepth

这是一个用于WideDepth的FiftyOne数据集。WideDepth是一个室内鱼眼深度估计基准（ICRA 2026），具有从高分辨率LiDAR扫描渲染的毫米精度地面真实深度和视差。数据集使用固定的相机配置：195°视场角、300毫米焦距、CENTER立体位置，覆盖101个室内场景。数据集包含398个样本，分组为101个场景，每个场景有最多四个切片：fisheye（鱼眼RGB和深度图像）、pano（等距柱面投影RGB、深度和视差图像）、pano_crop（裁剪的等距柱面投影图像，作为默认切片）和pointcloud（从pano_crop视图生成的彩色3D点云）。深度和视差存储为热图标签，以16位PNG格式提供，单位为毫米。数据集旨在支持深度估计和3D视觉任务的研究。This is a FiftyOne dataset for WideDepth — an indoor fisheye depth-estimation benchmark (ICRA 2026) with millimeter-accurate ground-truth depth and disparity rendered from high-resolution LiDAR scans. The dataset uses one fixed camera configuration (195° FOV, 300 mm focal length, CENTER stereo position) across 101 indoor scenes. It contains 398 total samples grouped into 101 scenes, with up to four slices per scene: fisheye (RGB and depth images), pano (equirectangular RGB, depth, and disparity images), pano_crop (cropped equirectangular images, as the default slice), and pointcloud (colored 3D point clouds generated from the pano_crop view). Depth and disparity are stored as heatmap labels in 16-bit PNG format with millimeter units. The dataset is designed for research in depth estimation and 3D vision tasks.