NBMOD (Noisy Background Multi-Object Dataset for grasp detection)
Introduced by Cao et al. in NBMOD: Find It and Grasp It in Noisy Background
Introduction
NBMOD is a dataset created for researching the task of specific object grasp detection by robots in noisy environments. The dataset comprises three subsets: Simple background Single-object Subset (SSS), Noisy background Single-object Subset (NSS), and Multi-Object grasp detection Subset (MOS). The SSS subset contains 13,500 images, the NSS subset contains 13,000 images, and the MOS subset contains 5,000 images.
What are the differences of NBMOD?
Unlike the renowned Cornell dataset, the NBMOD dataset differs in that its backgrounds are no longer simple whiteboards. The NSS and MOS subsets comprise a substantial number of images with noise, where this noise corresponds to interfering objects unrelated to the target objects for grasping detection. Moreover, in the MOS subset, each image encompasses multiple target objects for grasp detection, which closely resembles real-world working environments.
Papers
| Paper | Code | Results | Date | Stars |
|---|
