Leveraging Temporal Information for 3D Detection and Domain Adaptation

TL;DR

This paper introduces a simple method to incorporate temporal information via timestamps into LiDAR point cloud processing, significantly enhancing 3D detection performance in autonomous driving scenarios.

Contribution

It proposes a straightforward approach to utilize temporal data in point cloud learning, improving detection accuracy across multiple classes.

Findings

Consistent performance improvements across all three object classes.

Effective integration of temporal information with minimal modifications.

Enhanced 3D detection accuracy in autonomous driving datasets.

Abstract

Ever since the prevalent use of the LiDARs in autonomous driving, tremendous improvements have been made to the learning on the point clouds. However, recent progress largely focuses on detecting objects in a single 360-degree sweep, without extensively exploring the temporal information. In this report, we describe a simple way to pass such information in the learning pipeline by adding timestamps to the point clouds, which shows consistent improvements across all three classes.