Imitrob

Imitrob数据集是由捷克技术大学信息、机器人和控制论研究所创建的，专注于6D对象姿态估计的训练和评估。该数据集包含9种不同工具和12种操作任务的图像序列，涉及4个不同的演示者和两种摄像机视角。数据集中的每张图像都配有通过HTC Vive运动跟踪设备获得的精确6D对象姿态地面实况测量。Imitrob数据集特别适用于模仿学习和其他需要人类手持工具并执行任务的应用场景。该数据集的应用领域包括机器人学习、计算机视觉自动化和抓取操作感知，旨在解决对象姿态估计在遮挡和复杂环境中的挑战。

The Imitrob dataset was created by the Institute of Information, Robotics and Cybernetics at Czech Technical University, and focuses on the training and evaluation of 6D object pose estimation. This dataset comprises image sequences of 9 distinct tools and 12 manipulation tasks, involving 4 different demonstrators and two camera viewpoints. Each image in the dataset is paired with precise 6D object pose ground truth measurements obtained via HTC Vive motion tracking equipment. The Imitrob dataset is particularly suitable for imitation learning and other application scenarios where humans hold tools and perform tasks. Its application domains include robotic learning, computer vision automation, and grasping and manipulation perception, and it aims to address the challenges of object pose estimation under occlusion and complex environments.