CVGL Camera Calibration Dataset

数据集是使用 CARLA Simulator 的 Town 1 和 Town 2 生成的。该数据集由 48 美元的相机配置组成，每个城镇都有 24 美元的配置。为生成配置而修改的参数包括 $fov$、$x$、$y$、$z$、pitch、yaw 和 roll。这里，$fov$ 是视野，(x, y, z) 是平移，而 (pitch, yaw, and roll) 是摄像机之间的旋转。图像对的总数为 79,320 美元，其中 18,083 美元属于城镇 1，而 61,237 美元属于城镇 2，图像数量的差异是由于轨道的长度。

This dataset is generated using Town 1 and Town 2 from the CARLA Simulator. It consists of 48 camera configurations, with 24 configurations allocated to each town. The parameters modified for generating these configurations include $fov$, $x$, $y$, $z$, pitch, yaw, and roll. Here, $fov$ refers to the field of view, (x, y, z) represent the translation coordinates, and (pitch, yaw, roll) denote the camera's rotational parameters. The total number of image pairs is 79,320, of which 18,083 belong to Town 1 and 61,237 belong to Town 2. The discrepancy in the number of images arises from the difference in track lengths.

搜集汇总

背景与挑战

背景概述

该数据集基于CARLA Simulator在Town 1和Town 2生成，包含48种相机配置，总共有79,320对图像，用于相机校准相关研究。