Research map / Detection and segmentation / 3D detection and driving
Uni3D: A Unified Baseline for Multi-Dataset 3D Object Detection
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Train one detector jointly across Waymo, nuScenes, and KITTI using Uni3D rather than training on one dataset | Voxel-RCNN · Waymo+nuScenes+KITTI, car class average across datasets | 3D AP: 58.75 / 49.92 / 38.67 -> 66.89 | Helped |
| Align point-cloud coordinate ranges across datasets (ablation) | Voxel-RCNN multi-dataset training · Waymo-KITTI consolidation, Waymo validation | AP_3D: 26.93 -> 74.83 | Helped |
| Add statistics-level alignment to coordinate alignment (ablation) | Voxel-RCNN multi-dataset training · Waymo+nuScenes, Waymo validation, Vehicle | AP: 69.40 -> 75.16 | Helped |
| Add coupling-recoupling to coordinate alignment (ablation) | Voxel-RCNN multi-dataset training · Waymo+nuScenes, nuScenes validation, Car | AP_3D: 29.04 -> 42.25 | Helped |
| Use both statistics alignment and coupling-recoupling (ablation) | Voxel-RCNN multi-dataset training · KITTI+nuScenes, KITTI validation, Car | AP_3D: 82.68 / 82.50 -> 83.10 | Helped |
| Remove attention design from coupling-recoupling (ablation) | Uni3D · Waymo+nuScenes, Waymo validation, Vehicle | AP: 75.06 -> 75.26 | Hurt |
| Remove the SE component from coupling-recoupling (ablation) | Uni3D · Waymo+nuScenes, Waymo validation, Vehicle | AP: 74.37 -> 75.26 | Hurt |
| Copy the single-dataset BEV feature to both branches rather than mask the other branch (ablation) | Uni3D inference · Waymo+nuScenes, Waymo validation, Vehicle | AP: 73.78 -> 75.26 | Helped |
| Apply coordinate-origin alignment to KITTI-nuScenes consolidation (ablation) | Voxel-RCNN multi-dataset training · KITTI validation, Pedestrian and Cyclist | AP_3D: 53.67 -> 39.53; 42.41 -> 41.64 | Mixed |
Closest papers by meaning
- UniDet3D: Multi-dataset Indoor 3D Object Detection (AAAI 2025 · similarity 0.91)
- Uni$^2$Det: Unified and Universal Framework for Prompt-Guided Multi-dataset 3D Detection (ICLR 2025 · similarity 0.91)
- ImOV3D: Learning Open Vocabulary Point Clouds 3D Object Detection from Only 2D Images (NeurIPS 2024 · similarity 0.90)
- Training an Open-Vocabulary Monocular 3D Detection Model without 3D Data (NeurIPS 2024 · similarity 0.90)
- UNION: Unsupervised 3D Object Detection using Object Appearance-based Pseudo-Classes (NeurIPS 2024 · similarity 0.90)
- Unified Domain Generalization and Adaptation for Multi-View 3D Object Detection (NeurIPS 2024 · similarity 0.90)
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.