Abstract

The dataset contains the source code and model weights utilized for the experimental validation on segmentation of deformable linear objects. The developed approach is called FASTDLO. The source code algorithm comprises a deep convolutional neural network employed for background segmentation, the intersections between different Deformable Linear Objects (DLOs) are solved with a similarity-based network combined to a skeletonization algorithm. FASTDLO also describes each DLO instance with a sequence of 2D coordinates. The associated publication is the following: A. Caporali, K. Galassi, R. Zanella and G. Palli, "FASTDLO: Fast Deformable Linear Objects Instance Segmentation," in IEEE Robotics and Automation Letters, vol. 7, no. 4, pp. 9075-9082, Oct. 2022, doi: 10.1109/LRA.2022.3189791.