PCSCNet: Fast 3D semantic segmentation of LiDAR point cloud for autonomous car using point convolution and sparse convolution network

The autonomous car must recognize the driving environment quickly for safe driving. As Light Detection And Range (LiDAR) sensor is widely used in car, fast semantic segmentation of LiDAR point cloud, which point-wise classification cloud within framerate, has attracted attention recognition environment. Although voxel and fusion-based models are state-of-the-art model recently, their real-time performance suffer from high computational load due to resolution. In this paper, we propose voxel-based using Point Convolution 3D Sparse (PCSCNet). proposed designed outperform at both low resolution convolution-based feature extraction. Moreover, accelerates propagation sparse convolution after experimental results demonstrate that outperforms SemanticKITTI nuScenes, achieves inference.