混合现实手遮挡容器数据集

混合现实手遮挡容器数据集是由伦敦玛丽女王大学智能感知中心创建，旨在解决家庭容器6D姿态和尺寸估计问题。该数据集包含138,240张渲染的手和前臂持有48个合成对象的图像，分为3种抓取类别，跨越30个真实背景。数据集创建过程中，通过在真实背景上渲染合成手持对象，考虑场景中的位置和合理的照明，生成伪真实复合图像。该数据集主要应用于增强现实、在线购物、机器人抓取和操作等领域，以提高6D姿态和尺寸估计的准确性。

The Mixed Reality Hand-occluded Container Dataset was created by the Intelligent Sensing Center of Queen Mary University of London, aiming to address the 6D pose and size estimation problem for household containers. This dataset contains 138,240 rendered images of hands and forearms holding 48 synthetic objects, which are categorized into 3 grasping types and cover 30 real-world backgrounds. During the dataset creation process, pseudo-real composite images are generated by rendering synthetic handheld objects onto real backgrounds, taking into account the in-scene positions and reasonable lighting conditions. This dataset is mainly applied in fields such as augmented reality, online shopping, robotic grasping and manipulation to improve the accuracy of 6D pose and size estimation.