SUIT: Learning Significance-guided Information for 3D Temporal Detection

TL;DR

SUIT introduces a sparse, significance-guided approach for 3D temporal detection that efficiently fuses information across frames, reducing computational costs while maintaining high detection performance.

Contribution

The paper proposes a novel sparse feature-based temporal fusion method with significance-guided sampling and geometric transformation learning for 3D detection.

Findings

Reduces memory and computation costs significantly.

Achieves competitive or superior detection performance.

Effective in large-scale autonomous driving datasets.

Abstract

3D object detection from LiDAR point cloud is of critical importance for autonomous driving and robotics. While sequential point cloud has the potential to enhance 3D perception through temporal information, utilizing these temporal features effectively and efficiently remains a challenging problem. Based on the observation that the foreground information is sparsely distributed in LiDAR scenes, we believe sufficient knowledge can be provided by sparse format rather than dense maps. To this end, we propose to learn Significance-gUided Information for 3D Temporal detection (SUIT), which simplifies temporal information as sparse features for information fusion across frames. Specifically, we first introduce a significant sampling mechanism that extracts information-rich yet sparse features based on predicted object centroids. On top of that, we present an explicit geometric transformation learning technique, which learns the object-centric transformations among sparse features across frames. We evaluate our method on large-scale nuScenes and Waymo dataset, where our SUIT not only significantly reduces the memory and computation cost of temporal fusion, but also performs well over the state-of-the-art baselines.