BigHand2.2M
收藏资源简介:
BigHand2.2M数据集是由伦敦帝国学院创建的大规模手势数据集,包含220万张深度图像,每张图像都精确标注了21个关节位置。数据集通过使用六个6D磁性传感器和逆向运动学自动获取标注,旨在全面覆盖自然手势空间。该数据集不仅在数量上超越现有基准,而且在手势的多样性和标注质量上也有显著提升。BigHand2.2M数据集的应用领域包括手势识别和深度学习模型的训练,旨在解决现有数据集在数量、覆盖范围和标注精度上的限制问题。
The BigHand2.2M dataset is a large-scale gesture dataset developed by Imperial College London. It contains 2.2 million depth images, each with precise annotations of 21 joint positions. The annotations were automatically acquired using six 6D magnetic sensors and inverse kinematics, with the goal of comprehensively covering the natural gesture space. This dataset not only outperforms existing benchmarks in terms of scale, but also achieves significant improvements in gesture diversity and annotation quality. Applications of the BigHand2.2M dataset include gesture recognition and deep learning model training, aiming to address the limitations of existing datasets in terms of quantity, coverage and annotation accuracy.

- BigHand2.2M数据集首次发表,该数据集包含超过220万张手部图像,用于手部姿态估计研究。
- BigHand2.2M数据集首次应用于手部姿态估计算法的研究和开发,推动了相关领域的发展。
- 基于BigHand2.2M数据集的研究成果在多个国际会议上展示,进一步提升了该数据集的影响力。
- BigHand2.2M数据集被广泛应用于手部姿态估计的深度学习模型训练,成为该领域的重要基准数据集。
- BigHand2.2M数据集的扩展版本发布,增加了更多的手部姿态多样性和复杂性,进一步丰富了数据集的内容。
- 1BigHand2.2M: A Dataset for Estimating 3D Hand Pose from RGB ImagesUniversity of Edinburgh · 2020年
- 23D Hand Pose Estimation: A SurveyUniversity of Surrey · 2021年
- 3Hand Pose Estimation Using Deep Learning: A Comprehensive ReviewUniversity of California, Berkeley · 2022年
- 4A Comparative Study of 3D Hand Pose Estimation MethodsStanford University · 2021年
- 5Deep Learning for 3D Hand Pose Estimation: A Review and AnalysisMassachusetts Institute of Technology · 2022年



