CENet: Toward Concise and Efficient LiDAR Semantic Segmentation for Autonomous Driving

TL;DR

CENet is a novel, efficient LiDAR semantic segmentation network that enhances feature representation and reduces computational complexity, achieving superior accuracy and speed on benchmark datasets for autonomous driving.

Contribution

The paper introduces CENet, a concise and efficient architecture that combines larger kernel convolutions, specialized activation functions, and auxiliary heads for improved LiDAR segmentation.

Findings

CENet outperforms state-of-the-art models on SemanticKITTI and SemanticPOSS benchmarks.

It achieves higher mIoU with lower inference time.

The architecture effectively balances accuracy and computational efficiency.

Abstract

Accurate and fast scene understanding is one of the challenging task for autonomous driving, which requires to take full advantage of LiDAR point clouds for semantic segmentation. In this paper, we present a \textbf{concise} and \textbf{efficient} image-based semantic segmentation network, named \textbf{CENet}. In order to improve the descriptive power of learned features and reduce the computational as well as time complexity, our CENet integrates the convolution with larger kernel size instead of MLP, carefully-selected activation functions, and multiple auxiliary segmentation heads with corresponding loss functions into architecture. Quantitative and qualitative experiments conducted on publicly available benchmarks, SemanticKITTI and SemanticPOSS, demonstrate that our pipeline achieves much better mIoU and inference performance compared with state-of-the-art models. The code will be available at https://github.com/huixiancheng/CENet.