TFNet: Exploiting Temporal Cues for Fast and Accurate LiDAR Semantic Segmentation

TL;DR

TFNet leverages temporal cues and a max-voting post-processing method to improve the accuracy of range-image-based LiDAR semantic segmentation, effectively addressing occlusion issues in autonomous driving scenarios.

Contribution

The paper introduces a novel temporal fusion layer and a max-voting post-processing technique for range-image-based LiDAR segmentation, enhancing accuracy and robustness.

Findings

Improved segmentation accuracy on benchmark datasets.

Effective reduction of occlusion-related errors.

Post-processing method is versatile and applicable to various networks.

Abstract

LiDAR semantic segmentation plays a crucial role in enabling autonomous driving and robots to understand their surroundings accurately and robustly. A multitude of methods exist within this domain, including point-based, range-image-based, polar-coordinate-based, and hybrid strategies. Among these, range-image-based techniques have gained widespread adoption in practical applications due to their efficiency. However, they face a significant challenge known as the ``many-to-one'' problem caused by the range image's limited horizontal and vertical angular resolution. As a result, around 20% of the 3D points can be occluded. In this paper, we present TFNet, a range-image-based LiDAR semantic segmentation method that utilizes temporal information to address this issue. Specifically, we incorporate a temporal fusion layer to extract useful information from previous scans and integrate it with the current scan. We then design a max-voting-based post-processing technique to correct false predictions, particularly those caused by the ``many-to-one'' issue. We evaluated the approach on two benchmarks and demonstrated that the plug-in post-processing technique is generic and can be applied to various networks.