PDES-Net: LiDAR point cloud semantic segmentation network based on point-wise distance encoding and pointed-seg head

Engineering notes

The performance of the proposed PDES-Net is evaluated on the publicly available benchmarks, SemanticKITTI and nuScenes, achieving mIoU of 68.9% and 78.9%, respectively. Overall, the network achieves a well-balanced trade-off between accuracy and computational efficiency.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Abstract LiDAR-based semantic segmentation is significant in advanced autonomous driving systems. However, LiDAR point cloud data is sparse and unevenly distributed. This poses challenges to achieving accurate and efficient semantic understanding. To address the above problem, we propose a network model, PDES-Net. We design a point-wise distance encoding mechanism that introduces normalized continuous depth information as a supplementary feature into the network. The mechanism can enhance the perception of both near and distant points in the model and reduce geometric information loss. To enhance the expressive ability of point-wise prediction, we present a pointed-seg head module. This module adaptively integrates multilevel features through learnable weight coefficients. The performance of the proposed PDES-Net is evaluated on the publicly available benchmarks, SemanticKITTI and nuScenes, achieving mIoU of 68.9% and 78.9%, respectively. The proposed PDES-Net enhances segmentation performance while maintaining advantages in model parameters and inference speed. Overall, the network achieves a well-balanced trade-off between accuracy and computational efficiency.

Links and sources

Comments