Abstract: 3D point cloud object detection plays an important role in autonomous driving. However, labeling 3D object boxes is expensive and time-consuming, limiting the number of annotated point ...
Abstract: Recently, many self-supervised pre-training methods have been proposed to improve the performance of deep neural networks (DNNs) for 3D point clouds processing. However, the common mechanism ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results