HandSeg
收藏资源简介:
HandSeg数据集是由维多利亚大学创建,专注于从深度图像中进行手部分割的大规模数据集。该数据集包含210,000个条目,通过使用RGBD传感器和彩色手套自动生成高质量的手部分割标注。数据集的创建过程利用了颜色和深度通道来生成精确的地面实况标注,仅需用户佩戴彩色手套,无需复杂设备。HandSeg数据集广泛应用于实时手部分割算法的训练,旨在解决现有数据集在区分双手方面的不足,提高算法的准确性和鲁棒性。
The HandSeg dataset is a large-scale dataset developed by the University of Victoria, focusing on hand segmentation from depth images. It contains 210,000 entries, and automatically generates high-quality hand segmentation annotations using RGBD sensors and colored gloves. The dataset creation process leverages color and depth channels to produce precise ground-truth annotations, which only requires users to wear colored gloves without the need for complex equipment. The HandSeg dataset is widely used for training real-time hand segmentation algorithms, aiming to address the shortcomings of existing datasets in distinguishing between two hands and improve the accuracy and robustness of the algorithms.

- 1HandSeg: An Automatically Labeled Dataset for Hand Segmentation from Depth Images维多利亚大学 · 2018年



