OcTr: Octree-based Transformer for 3D Object Detection

TL;DR

OcTr introduces an octree-based Transformer for 3D object detection that efficiently captures global context and improves accuracy in large-scale LiDAR scenes, especially for distant and occluded objects.

Contribution

The paper proposes a novel octree-based Transformer architecture with a hybrid positional embedding for enhanced 3D object detection.

Findings

Achieves state-of-the-art results on Waymo and KITTI datasets.

Effectively balances accuracy and computational efficiency.

Captures rich global context through a coarse-to-fine octree structure.

Abstract

A key challenge for LiDAR-based 3D object detection is to capture sufficient features from large scale 3D scenes especially for distant or/and occluded objects. Albeit recent efforts made by Transformers with the long sequence modeling capability, they fail to properly balance the accuracy and efficiency, suffering from inadequate receptive fields or coarse-grained holistic correlations. In this paper, we propose an Octree-based Transformer, named OcTr, to address this issue. It first constructs a dynamic octree on the hierarchical feature pyramid through conducting self-attention on the top level and then recursively propagates to the level below restricted by the octants, which captures rich global context in a coarse-to-fine manner while maintaining the computational complexity under control. Furthermore, for enhanced foreground perception, we propose a hybrid positional embedding, composed of the semantic-aware positional embedding and attention mask, to fully exploit semantic and geometry clues. Extensive experiments are conducted on the Waymo Open Dataset and KITTI Dataset, and OcTr reaches newly state-of-the-art results.