TUM VI benchmark

TUM VI benchmark是由慕尼黑工业大学创建的一个新型数据集，专门用于评估视觉惯性测程（VI odometry）。该数据集包含多种场景下的序列，提供20Hz、1024x1024分辨率的相机图像，以及200Hz的IMU测量数据，所有传感器通过硬件实现时间同步。数据集还包括从运动捕捉系统获取的高频（120Hz）精确姿态地面实况，用于轨迹评估。此数据集适用于增强现实和机器人技术等领域，旨在解决视觉惯性测程方法的评估和优化问题。

The TUM VI benchmark is a novel dataset created by the Technical University of Munich (TUM) specifically for evaluating visual-inertial odometry (VI odometry). This dataset contains sequences across various scenarios, providing camera images with a resolution of 1024x1024 at 20Hz, as well as IMU measurement data at 200Hz. All sensors are time-synchronized via hardware. The dataset also includes high-frequency (120Hz) accurate pose ground truth obtained from a motion capture system, which is used for trajectory evaluation. This dataset is applicable to fields such as augmented reality and robotics, aiming to address the evaluation and optimization issues of visual-inertial odometry methods.