Abstract

This dataset contains experimental data related to robotic grasping applications, produced in the framework of REMODEL project. Specifically, it contains the collection of 15 tabletop scenes, with 12 randomly arranged objects, featuring different backgrounds: 3 scenes with homogeneous background; 3 scenes with wood; 3 scenes with black background; and 5 scenes with a high-clutter background (several prints of Pollock’s painting). The data are presented in the publication: De Gregorio, D., Zanella, R., Palli, G., & Di Stefano, L. (2020). Effective Deployment of CNNs for 3DoF Pose Estimation and Grasping in Industrial Settings. (in press).