PEGASUS, Ramen Dataset, PEGASET Dataset

PEGASUS是一个基于3D高斯溅射的通用数据集生成器，用于6DOF物体姿态估计。该数据集通过使用普通相机获取环境和物体表示，利用高斯溅射重建技术，结合物理引擎模拟自然物体放置，生成大量新场景。数据集包含RGB图像、深度图、语义掩码和6DOF物体姿态等多种数据点。该数据集的应用领域包括机器人技术，旨在通过合成数据提高真实世界数据中的姿态估计准确性。

PEGASUS is a general-purpose dataset generator based on 3D Gaussian Splatting, tailored for 6DOF object pose estimation. It acquires environment and object representations using conventional cameras, leverages 3D Gaussian Splatting reconstruction techniques, and integrates with physics engines to simulate natural object placement, thereby generating a vast number of novel scenes. The dataset encompasses diverse data modalities including RGB images, depth maps, semantic masks, and 6DOF object poses. Its application scenarios cover robotics, with the core objective of enhancing the accuracy of pose estimation in real-world data via synthetic data.