RaCFormer: Towards High-Quality 3D Object Detection via Query-based Radar-Camera Fusion

TL;DR

RaCFormer is a novel query-based radar-camera fusion transformer that significantly improves 3D object detection accuracy by adaptively sampling features and leveraging radar Doppler data, achieving state-of-the-art results.

Contribution

The paper introduces a query-based fusion framework with adaptive sampling, radar-guided depth refinement, and temporal modeling for enhanced 3D detection.

Findings

Achieves 64.9% mAP on nuScenes

Sets new state-of-the-art on VoD dataset

Demonstrates effective radar and camera feature integration

Abstract

We propose Radar-Camera fusion transformer (RaCFormer) to boost the accuracy of 3D object detection by the following insight. The Radar-Camera fusion in outdoor 3D scene perception is capped by the image-to-BEV transformation--if the depth of pixels is not accurately estimated, the naive combination of BEV features actually integrates unaligned visual content. To avoid this problem, we propose a query-based framework that enables adaptive sampling of instance-relevant features from both the bird's-eye view (BEV) and the original image view. Furthermore, we enhance system performance by two key designs: optimizing query initialization and strengthening the representational capacity of BEV. For the former, we introduce an adaptive circular distribution in polar coordinates to refine the initialization of object queries, allowing for a distance-based adjustment of query density. For the latter, we initially incorporate a radar-guided depth head to refine the transformation from image view to BEV. Subsequently, we focus on leveraging the Doppler effect of radar and introduce an implicit dynamic catcher to capture the temporal elements within the BEV. Extensive experiments on nuScenes and View-of-Delft (VoD) datasets validate the merits of our design. Remarkably, our method achieves superior results of 64.9% mAP and 70.2% NDS on nuScenes. RaCFormer also secures the state-of-the-art performance on the VoD dataset. Code is available at https://github.com/cxmomo/RaCFormer.