AI Paper Atlas

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

RPGFusion: 4D Radar Prior-Guided Multi-Modal Fusion for 3D Detection

Xin Qiu, Wenjie Liu

CVPR 2026 · Detection and segmentation › 3D detection and driving · approach: Camera-radar fusion detection

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 radar confidence and depth prior maps to image BEV query initialization (ablation)RPGFusion · VoD val, entire annotated areamAP: 58.29 -> 69.31Improved
add radar confidence and depth prior maps to image-to-BEV sampling (ablation)RPGFusion · VoD val, entire annotated areamAP: 52.47 -> 69.31Improved
add depth map to image sampling (vs confidence map only) (ablation)RPGFusion · VoD val, entire annotated areamAP: 57.60 -> 69.31 without depth, 63.46 without confidenceImproved
add radar robust encoding (mean/median stats and RCS confidence) (ablation)RPGFusion · VoD val, entire annotated areamAP: 66.68 -> 69.31Improved
add sparse-to-dense radar densification (ablation)RPGFusion · VoD val, entire annotated areamAP: 59.34 -> 69.31Improved
add spatial alignment for both camera and radar (ablation)RPGFusion · VoD val, entire annotated areamAP: drop of 7.05 mAP when removedImproved
add bidirectional semantic fusion between camera and radar (ablation)RPGFusion · VoD val, entire annotated areamAP: drop of 16.52 mAP when removedImproved
replace concatenation with gated unified fusion (ablation)RPGFusion · VoD val, entire annotated areamAP: 63.25 -> 69.31Improved
add radar to camera-only model (fusion) (ablation)RPGFusion · VoD val, entire annotated areamAP: 56.09 -> 69.31Improved
add camera to radar-only model (fusion) (ablation)RPGFusion · VoD val, entire annotated areamAP: 45.71 -> 69.31Improved
use RPGFusion instead of CVFusionCVFusion · VoD val, entire annotated areamAP: 65.41 -> 69.31Improved
use RPGFusion instead of CVFusionCVFusion · TJ4DRadSet test, 3DmAP: 40.00 -> 43.05Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 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