Ego2HandsPose
收藏资源简介:
Ego2HandsPose是一个专为单目RGB相机在非实验室环境下进行双人手3D全局姿态估计而设计的大型数据集。该数据集由杨百翰大学的研究团队创建,旨在解决现有数据集在视觉多样性和环境适应性方面的不足。数据集通过一种基于合成的数据生成技术,创建了具有高质量、数量和多样性的双人手实例,这些实例能够很好地泛化到未见过的领域。Ego2HandsPose不仅支持双人手分割和检测,还首次实现了在未知环境中的彩色双人手3D跟踪。该数据集的应用领域包括人机交互、手势识别和虚拟现实/增强现实/混合现实等,旨在提高这些应用中的用户体验和交互的自然性。
Ego2HandsPose is a large-scale dataset specifically designed for 3D global pose estimation of two-person hands using monocular RGB cameras in unconstrained non-laboratory environments. Developed by the research team at Brigham Young University, it aims to address the limitations of existing datasets in terms of visual diversity and environmental adaptability. The dataset generates high-quality, plentiful and diverse two-person hand instances through a synthetic data generation approach, which enables strong generalization to unseen domains. Ego2HandsPose not only supports two-person hand segmentation and detection, but also achieves, for the first time, color-aware 3D tracking of two-person hands in unknown environments. Its application scenarios include human-computer interaction, gesture recognition, virtual reality (VR), augmented reality (AR) and mixed reality (MR), with the goal of enhancing user experience and the naturalness of interaction in these applications.

- 1Ego2HandsPose: A Dataset for Egocentric Two-hand 3D Global Pose Estimation杨百翰大学 · 2022年



