Self-ROBI
收藏资源简介:
Self-ROBI数据集是由新加坡国立大学机械工程系提出的一个用于反射对象位姿估计自训练的机器人抓取数据集。该数据集通过结合高分辨率相机捕获的重建数据和低成本相机捕获的实时数据,使用多对象位姿推理算法进行优化,生成用于自训练的伪标签。数据集旨在解决低成本相机在抓取反射对象时遇到的挑战,如稀疏深度信息和变化的反射纹理。
The Self-ROBI dataset is a robotic grasping dataset for self-training in reflective object pose estimation, proposed by the Department of Mechanical Engineering at the National University of Singapore. This dataset combines reconstructed data captured by high-resolution cameras and real-time data captured by low-cost cameras, uses multi-object pose inference algorithms for optimization, and generates pseudo-labels for self-training. The dataset aims to address the challenges faced when using low-cost cameras to grasp reflective objects, such as sparse depth information and varying reflective textures.

- 1Reasoning and Learning a Perceptual Metric for Self-Training of Reflective Objects in Bin-Picking with a Low-cost Camera新加坡国立大学机械工程系 · 2025年



