AI Paper Atlas

Research map / Detection and segmentation / 3D detection and driving

RayD3D: Distilling Depth Knowledge Along the Ray for Robust Multi-View 3D Object Detection

Rui Ding, Zhaonian Kuang, Zongwei Zhou, Meng Yang, Xinhu Zheng, Gang Hua

AAAI 2026 · Detection and segmentation › 3D detection and driving · approach: Multi-camera BEV detection · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add RCD module to BEVDet (ablation)BEVDet · BEVDet nuScenes val / RoboBEVNDS/mAP: 37.4/29.6 -> 40.6/31.8Improved
add RCD module to BEVDet (ablation)BEVDet · BEVDet nuScenes val / RoboBEVmRR: 52.2 -> 53.7Improved
add RWD module to BEVDet (ablation)BEVDet · BEVDet nuScenes val / RoboBEVNDS/mAP: 37.4/29.6 -> 41.7/32.0Improved
add RWD module to BEVDet (ablation)BEVDet · BEVDet nuScenes val / RoboBEVmRR: 52.2 -> 54.1Improved
add RWD on top of RCD (ablation)BEVDet + RCD · BEVDet nuScenes val / RoboBEVNDS/mAP: 40.6/31.8 -> 41.9/32.7Improved
add RCD on top of RWD (ablation)BEVDet + RWD · BEVDet nuScenes val / RoboBEVmRR: 54.1 -> 54.7Improved
sample negatives with Gaussian instead of random (ablation)RCD · BEVDet nuScenes val / RoboBEVNDS/mAP: 39.4/29.9 -> 40.6/31.8Improved
also sample negatives in the teacher network (Gaussian) (ablation)RCD · BEVDet nuScenes val / RoboBEVmRR: 53.0 -> 53.7Improved
use KL divergence instead of cosine similarity for RWD weights (ablation)RWD · BEVDet nuScenes val / RoboBEVmRR: 53.2 -> 54.1Improved
use KL divergence instead of JS divergence for RWD weights (ablation)RWD · BEVDet nuScenes val / RoboBEVNDS/mAP: 41.0/31.8 -> 41.7/32.0Improved
add RayD3D to BEVFormerBEVFormer · nuScenes val / RoboBEVmRR: 62.1 -> 64.8Improved
add RayD3D to BEVDepth4D-r101 (Snow corruption)BEVDepth4D-r101 · RoboBEV SnowNDS ratio (%): 18.5 -> 22.1Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy