Jacquard grasping dataset
收藏资源简介:
Jacquard grasping dataset是由里昂中央理工学院LIRIS实验室创建的大规模合成数据集,基于ShapeNet数据集的子集ShapeNetSem构建。该数据集包含超过50000张RGB-D图像,涵盖11000多个不同对象,每个对象有多个标注的抓取位置。创建过程中,利用pyBullet和Blender软件模拟真实环境,生成高质量的图像和抓取位置标注。Jacquard数据集主要用于训练深度神经网络,以预测机器人抓取位置,特别适用于未见过的对象,旨在提高机器人在实际应用中的抓取技能。
The Jacquard Grasping Dataset is a large-scale synthetic dataset created by the LIRIS Laboratory at École Centrale de Lyon, built based on ShapeNetSem, a subset of the ShapeNet dataset. This dataset contains over 50,000 RGB-D images, covering more than 11,000 distinct objects, with multiple annotated grasp positions for each object. During its creation, pyBullet and Blender software were used to simulate real-world environments, generating high-quality images and grasp position annotations. The Jacquard Dataset is primarily used to train deep neural networks for predicting robot grasp positions, and is particularly suitable for unseen objects, aiming to improve the grasping capabilities of robots in real-world applications.




