YCB-Video Dataset

YCB-Video数据集是由华盛顿大学机器人与状态估计实验室开发的大规模视频数据集，专门用于6D物体姿态估计研究。该数据集包含21个YCB物体的92个视频序列，总计133,827帧RGB-D图像，提供了精确的6D姿态标注，能够有效处理遮挡和对称物体等复杂场景下的姿态估计问题，是计算机视觉和机器人领域的重要基准数据集。YCB-Video Dataset is a large-scale video dataset developed by the Robotics and State Estimation Lab at the University of Washington, specifically designed for 6D object pose estimation research. This dataset contains 92 video sequences encompassing 21 YCB objects, with a total of 133,827 RGB-D image frames, and provides accurate 6D pose annotations. It can effectively handle pose estimation problems in complex scenarios such as occlusion and symmetric objects, and is an important benchmark dataset in the fields of computer vision and robotics.

搜集汇总

背景与挑战

背景概述

YCB-Video Dataset是一个由华盛顿大学开发的大规模视频数据集，专门用于6D物体姿态估计研究，包含21个物体的92个视频序列和超过13万帧RGB-D图像，提供精确姿态标注，能有效处理遮挡和对称物体等复杂场景，是计算机视觉和机器人领域的重要基准数据集。