Transformers for Object Detection in Large Point Clouds

TL;DR

This paper introduces TransLPC, a transformer-based model designed for large-scale point cloud object detection, featuring a novel query refinement technique that enhances accuracy while managing computational resources effectively.

Contribution

The paper proposes a modified transformer architecture for large point clouds and introduces a query refinement method that improves detection accuracy without increasing memory usage.

Findings

Achieves improved detection accuracy on nuScenes dataset

Enables transformer models to handle larger point cloud inputs

Compatible with existing transformer-based detection and tracking methods

Abstract

We present TransLPC, a novel detection model for large point clouds that is based on a transformer architecture. While object detection with transformers has been an active field of research, it has proved difficult to apply such models to point clouds that span a large area, e.g. those that are common in autonomous driving, with lidar or radar data. TransLPC is able to remedy these issues: The structure of the transformer model is modified to allow for larger input sequence lengths, which are sufficient for large point clouds. Besides this, we propose a novel query refinement technique to improve detection accuracy, while retaining a memory-friendly number of transformer decoder queries. The queries are repositioned between layers, moving them closer to the bounding box they are estimating, in an efficient manner. This simple technique has a significant effect on detection accuracy, which is evaluated on the challenging nuScenes dataset on real-world lidar data. Besides this, the proposed method is compatible with existing transformer-based solutions that require object detection, e.g. for joint multi-object tracking and detection, and enables them to be used in conjunction with large point clouds.