LiDAR-based Recurrent 3D Semantic Segmentation with Temporal Memory Alignment

TL;DR

This paper introduces a recurrent 3D semantic segmentation method for LiDAR point clouds that leverages temporal memory alignment using ego motion, significantly improving segmentation accuracy over single-frame approaches.

Contribution

It proposes a novel recurrent architecture with Temporal Memory Alignment for 3D semantic segmentation, exploiting temporal information in LiDAR sequences for the first time.

Findings

Ranks first on SemanticKITTI multiple scan benchmark.

Achieves state-of-the-art on single scan benchmark.

Temporal information significantly improves segmentation results.

Abstract

Understanding and interpreting a 3d environment is a key challenge for autonomous vehicles. Semantic segmentation of 3d point clouds combines 3d information with semantics and thereby provides a valuable contribution to this task. In many real-world applications, point clouds are generated by lidar sensors in a consecutive fashion. Working with a time series instead of single and independent frames enables the exploitation of temporal information. We therefore propose a recurrent segmentation architecture (RNN), which takes a single range image frame as input and exploits recursively aggregated temporal information. An alignment strategy, which we call Temporal Memory Alignment, uses ego motion to temporally align the memory between consecutive frames in feature space. A Residual Network and ConvGRU are investigated for the memory update. We demonstrate the benefits of the presented approach on two large-scale datasets and compare it to several stateof-the-art methods. Our approach ranks first on the SemanticKITTI multiple scan benchmark and achieves state-of-the-art performance on the single scan benchmark. In addition, the evaluation shows that the exploitation of temporal information significantly improves segmentation results compared to a single frame approach.