EgoPressure
收藏资源简介:
EgoPressure数据集由苏黎世联邦理工学院创建,专注于从第一人称视角进行手部压力和姿态估计。该数据集包含5小时的触碰和压力交互记录,由21名参与者在移动的第一人称摄像头和7个静止的Kinect摄像头下完成。数据集通过8个同步的Azure Kinect摄像头捕捉RGB图像和深度图,并使用Sensel触摸传感器获取精确的接触和压力数据。创建过程包括多视角序列优化方法来标注手部姿态,确保数据的高质量。该数据集主要应用于增强现实(AR)、虚拟现实(VR)和机器人研究,旨在提高对手部与物体交互的物理理解。
The EgoPressure dataset was developed by ETH Zurich, focusing on hand pressure and pose estimation from a first-person perspective. This dataset contains five hours of touch and pressure interaction recordings, collected from twenty-one participants using a mobile first-person camera and seven stationary Kinect cameras. It captures RGB images and depth maps via eight synchronized Azure Kinect cameras, and acquires precise contact and pressure data with Sensel touch sensors. The dataset construction adopts multi-view sequence optimization techniques for hand pose annotation, ensuring the high quality of the collected data. This dataset is primarily applied to augmented reality (AR), virtual reality (VR), and robotics research, with the goal of advancing the physical understanding of hand-object interactions.

- 1EgoPressure: A Dataset for Hand Pressure and Pose Estimation in Egocentric Vision苏黎世联邦理工学院 · 2024年



