Row-wise LiDAR Lane Detection Network with Lane Correlation Refinement

TL;DR

This paper introduces a two-stage LiDAR-based lane detection network that leverages row-wise detection and attention mechanisms, achieving state-of-the-art accuracy with reduced computational cost and improved robustness to occlusions in autonomous driving scenarios.

Contribution

The paper presents a novel two-stage LiDAR lane detection network with a row-wise approach and attention-based refinement, outperforming existing methods in accuracy and efficiency.

Findings

Achieves 30% higher F1-score with less GFLOPs

Demonstrates robustness to lane occlusions

Outperforms state-of-the-art in K-Lane dataset

Abstract

Lane detection is one of the most important functions for autonomous driving. In recent years, deep learning-based lane detection networks with RGB camera images have shown promising performance. However, camera-based methods are inherently vulnerable to adverse lighting conditions such as poor or dazzling lighting. Unlike camera, LiDAR sensor is robust to the lighting conditions. In this work, we propose a novel two-stage LiDAR lane detection network with row-wise detection approach. The first-stage network produces lane proposals through a global feature correlator backbone and a row-wise detection head. Meanwhile, the second-stage network refines the feature map of the first-stage network via attention-based mechanism between the local features around the lane proposals, and outputs a set of new lane proposals. Experimental results on the K-Lane dataset show that the proposed network advances the state-of-the-art in terms of F1-score with 30% less GFLOPs. In addition, the second-stage network is found to be especially robust to lane occlusions, thus, demonstrating the robustness of the proposed network for driving in crowded environments.