INTELLIMAN. WP5. Grasping, Manipulation and Arm-Hand Coordination. T5_1. Data Fusion and Sensing Technology. Efficient Deformable Objects Labeling. v0

The dataset contains the user data collected during an experimental validation of a novel methodology for dataset labeling. The proposed method allows to efficiently label Deformable Objects (DOs) in images at the pixel level, starting from sparse annotations of key points. This allows the generation of a real-world dataset of DO images for segmentation purposes with minimal human effort. The approach comprises three main steps. First, a set of images is collected by a camera-equipped robotic arm. Second, a user performs sparse annotation via key points on just one image from the collected set. Third, the initial sparse annotations are converted into dense labels ready for segmentation tasks by leveraging a foundation model in zero-shot settings. The data were produced in the framework of Horizon Europe IntelliMan project and were presented in the following publication:
A. Caporali, K. Galassi, M. Pantano and G. Palli, "Deformable Objects Perception is Just a Few Clicks Away – Dense Annotations from Sparse Inputs", 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Abu Dhabi, United Arab Emirates, 2024, pp. 5880-5887, doi: 10.1109/IROS58592.2024.10802495.