SurgRIPE
收藏资源简介:
SurgRIPE数据集由伦敦帝国理工学院哈姆林机器人手术中心创建,旨在为手术机器人器械姿态估计提供真实的手术视频数据。该数据集包含2841条标注帧,涵盖了手术器械的3D姿态和分割掩码信息,数据来源于达芬奇手术系统的内窥镜视频。数据集的创建过程结合了基于标记的姿态估计和深度学习图像修复技术,生成了高精度的6自由度姿态数据。该数据集主要用于评估无标记姿态估计方法,推动机器人手术系统的精确性和自主性发展。
SurgRIPE dataset was created by the Hamlyn Centre for Robotic Surgery at Imperial College London, aiming to provide real-world surgical video data for robotic surgical instrument pose estimation. This dataset contains 2841 annotated frames covering 3D poses and segmentation masks of surgical instruments, with data sourced from endoscopic videos of the Da Vinci Surgical System. The dataset's creation process combines marker-based pose estimation and deep learning image inpainting techniques to generate high-precision 6-degree-of-freedom (6DoF) pose data. This dataset is primarily used to evaluate marker-free pose estimation methods, advancing the precision and autonomy of robotic surgical systems.

- 1SurgRIPE challenge: Benchmark of Surgical Robot Instrument Pose Estimation伦敦帝国理工学院哈姆林机器人手术中心 · 2025年



