Research map / Detection and segmentation / 3D detection and driving
Query-based Temporal Fusion with Explicit Motion for 3D Object Detection
Jinghua Hou, Zhe Liu, dingkang liang, Zhikang Zou, Xiaoqing Ye, Xiang Bai
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Add cross-attention temporal fusion (ablation) | TransFusion-L · nuScenes val | mAP: 65.0 -> 65.2 | No clear effect |
| Add Motion-guided Temporal Modeling (MTM) (ablation) | TransFusion-L · nuScenes val | mAP: 65.0 -> 66.2 | Improved |
| Add MTM (ablation) | TransFusion-L · nuScenes val | NDS: 70.0 -> 70.5 | Improved |
| Add IoU branch to rectify confidence (ablation) | QTNet · nuScenes val | mAP: 66.2 -> 66.3 | Improved |
| Decouple classification and regression temporal fusion (ablation) | QTNet · nuScenes val | NDS: 70.6 -> 70.8 | Improved |
| Increase frames from 1 to 2 (ablation) | QTNet on TransFusion-L · nuScenes val | mAP: 65.0 -> 66.1 | Improved |
| Increase frames from 3 to 4 (ablation) | QTNet on TransFusion-L · nuScenes val | mAP: 66.4 -> 66.5 | Improved |
| Increase frames from 4 to 5 (ablation) | QTNet on TransFusion-L · nuScenes val | mAP: 66.5 -> 66.4 | No clear effect |
| Increase frames from 1 to 5 (ablation) | QTNet on TransFusion-L · nuScenes val | latency (ms): +7.7 | Hurt |
| Query-based fusion instead of BEV-based MGTANet (ablation) | TransFusion-L · nuScenes val | mAP: 64.0 -> 64.7 | Improved |
| Query-based fusion instead of proposal-based MPPNet (ablation) | TransFusion-L · nuScenes val | mAP: 63.4 -> 64.7 | Improved |
| Query-based fusion instead of proposal-based MPPNet (ablation) | TransFusion-L · nuScenes val | latency overhead (ms): +127.9 -> +4.5 | Improved |
Closest papers by meaning
- Horizon3D: Sparse Radar-Camera Fusion for Long-Range 3D Perception in Autonomous Driving (ECCV 2026 · similarity 0.88)
- CRT-Fusion: Camera, Radar, Temporal Fusion Using Motion Information for 3D Object Detection (NeurIPS 2024 · similarity 0.88)
- PTT: Point-Trajectory Transformer for Efficient Temporal 3D Object Detection (CVPR 2024 · similarity 0.87)
- RaCFormer: Towards High-Quality 3D Object Detection via Query-based Radar-Camera Fusion (CVPR 2025 · similarity 0.86)
- Ev-3DOD: Pushing the Temporal Boundaries of 3D Object Detection with Event Cameras (CVPR 2025 · similarity 0.85)
- IS-Fusion: Instance-Scene Collaborative Fusion for Multimodal 3D Object Detection (CVPR 2024 · similarity 0.85)
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.
