HaDR
收藏资源简介:
HaDR数据集是由奥斯特拉瓦技术大学机械工程学院机器人系创建的合成RGB-D数据集,包含117,000个样本,用于训练多模态实例分割模型。该数据集通过领域随机化技术生成,涵盖了随机位置和方向的手部模型,以及各种几何形状和干扰物,以模拟复杂的工业环境。数据集的创建旨在解决现有数据集在颜色和手部位置上的局限性,特别适用于工业环境中的人机交互接口,以提高模型在复杂和非结构化环境中的预测准确性。
The HaDR dataset is a synthetic RGB-D dataset created by the Department of Robotics, Faculty of Mechanical Engineering, VŠB-Technical University of Ostrava, containing 117,000 samples intended for training multimodal instance segmentation models. Generated via domain randomization techniques, this dataset encompasses hand models with random positions and orientations, along with various geometric shapes and distractors, aiming to simulate complex industrial environments. The dataset was developed to address the limitations of existing datasets regarding color and hand positioning, and is particularly suitable for human-robot interaction interfaces in industrial environments, with the goal of improving the prediction accuracy of models in complex and unstructured environments.




