Omni6DPose
收藏资源简介:
Omni6DPose是一个全面的6D物体姿态估计数据集,由北京大学前沿计算研究中心创建。该数据集包含149个类别的581个实例,总计807000张图像,涵盖了广泛的物体材料和形状。数据集分为ROPE(真实6D物体姿态估计数据集)和SOPE(模拟6D物体姿态估计数据集)两个部分,分别包含332K和475K张图像。Omni6DPose的创建过程涉及高精度扫描和手动对齐,以确保数据的准确性和多样性。该数据集主要应用于机器人和增强现实/虚拟现实领域,旨在解决物体姿态估计和跟踪的问题。
Omni6DPose is a comprehensive 6D object pose estimation dataset developed by the Center for Frontiers of Computing Studies, Peking University. This dataset contains 581 instances across 149 categories, with a total of 807,000 images covering a wide range of object materials and shapes. It is divided into two subsets: ROPE (Real 6D Object Pose Estimation Dataset) and SOPE (Simulated 6D Object Pose Estimation Dataset), which hold 332K and 475K images respectively. The construction of Omni6DPose involves high-precision scanning and manual alignment to ensure the accuracy and diversity of the dataset. This dataset is primarily applied in robotics and augmented reality/virtual reality fields, aiming to address the challenges of object pose estimation and tracking.

- 1Omni6DPose: A Benchmark and Model for Universal 6D Object Pose Estimation and Tracking北京大学前沿计算研究中心 · 2024年



