A Convolutional Neural Network for Point Cloud Instance Segmentation in Cluttered Scene Trained by Synthetic Data Without Color

no code implementations • IEEE Access 2020 • Yajun Xu, Shogo Arai, Fuyuki Tokuda, Kazuhiro Kosuge

A great convergence of training loss often requires a large amount of human-annotated data, while making such a 3D dataset is time-consuming.

3D Instance Segmentation Autonomous Driving +2