SynMotor

SynMotor数据集由卡尔斯鲁厄理工学院人类与机器人研究所创建，包含1000个电动机模型的2D合成图像和3D合成点云数据。该数据集用于支持计算机视觉任务，如2D/3D检测、分类、分割及多属性学习。数据集中的电动机模型具有多种可学习属性，这些属性被量化为连续变量，适用于未充分探索的回归任务。此外，数据集还提供了针对每个任务的适当评估指标和基准结果，旨在推动对象属性学习和多任务学习领域的研究。

The SynMotor dataset was developed by the Institute of Human-Robot Systems, Karlsruhe Institute of Technology. It contains 2D synthetic images and 3D synthetic point cloud data for 1000 motor models. This dataset is intended to support computer vision tasks such as 2D/3D detection, classification, segmentation, and multi-attribute learning. The motor models included in the dataset feature multiple learnable attributes, which are quantified as continuous variables and applicable to under-explored regression tasks. Furthermore, the dataset provides appropriate evaluation metrics and baseline results for each task, with the goal of advancing research in the fields of object attribute learning and multi-task learning.