Ret3D: Rethinking Object Relations for Efficient 3D Object Detection in Driving Scenes

TL;DR

Ret3D introduces a novel two-stage LiDAR-based 3D object detection framework that effectively captures spatial and temporal object relations using relation modules, achieving state-of-the-art performance with minimal overhead.

Contribution

The paper proposes intra-frame and inter-frame relation modules integrated into a two-stage detector to exploit object relations efficiently in 3D detection.

Findings

Achieves 5.5% higher LEVEL 1 mAPH on WOD

Achieves 3.2% higher LEVEL 2 mAPH on WOD

State-of-the-art performance with negligible overhead

Abstract

Current efficient LiDAR-based detection frameworks are lacking in exploiting object relations, which naturally present in both spatial and temporal manners. To this end, we introduce a simple, efficient, and effective two-stage detector, termed as Ret3D. At the core of Ret3D is the utilization of novel intra-frame and inter-frame relation modules to capture the spatial and temporal relations accordingly. More Specifically, intra-frame relation module (IntraRM) encapsulates the intra-frame objects into a sparse graph and thus allows us to refine the object features through efficient message passing. On the other hand, inter-frame relation module (InterRM) densely connects each object in its corresponding tracked sequences dynamically, and leverages such temporal information to further enhance its representations efficiently through a lightweight transformer network. We instantiate our novel designs of IntraRM and InterRM with general center-based or anchor-based detectors and evaluate them on Waymo Open Dataset (WOD). With negligible extra overhead, Ret3D achieves the state-of-the-art performance, being 5.5% and 3.2% higher than the recent competitor in terms of the LEVEL 1 and LEVEL 2 mAPH metrics on vehicle detection, respectively.