On the Impact of LiDAR Point Cloud Compression on Remote Semantic Segmentation

Abstract: Autonomous vehicles rely on LiDAR sensors to generate 3D point clouds for accurate segmentation and object detection. In a context of a smart city framework, we would like to understand the effect that transmission (compression) can have on remote (cloud) segmentation, instead of local processing. In this short paper, we try to understand the impact of point cloud compression on semantic segmentation performance and to estimate the necessary bandwidth requirements. We developed a new (suitable) distortion metric to evaluate such an impact. Two of MPEG's compression algorithms (GPCC and L3C2) and two leading semantic segmentation algorithms (2DPASS and PVKD) were tested over the Semantic KITTI dataset. Results indicate that high segmentation quality requires communication throughput of approximately 0.6 MB/s for G-PCC and 2.8 MB/s for L3C2. These results are important in order to plan infrastructure resources for autonomous navigation.