Research map / Detection and segmentation / 3D detection and driving
3D Copy-Paste: Physically Plausible Object Insertion for Monocular 3D Detection
Yunhao Ge, Hong-Xing Yu, Cheng Zhao, Yuliang Guo, Xinyu Huang, Liu Ren, Laurent Itti, Jiajun Wu
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 |
|---|---|---|---|
| insert objects at random positions with camera point light (ablation) | ImVoxelNet · SUN RGB-D | mAP@0.25: 40.96 -> 37.02 | Hurt |
| add physically plausible position, size and pose (camera point light) (ablation) | ImVoxelNet · SUN RGB-D | mAP@0.25: 40.96 -> 41.80 | Improved |
| add plausible dynamic lighting and shadows on top of plausible geometry (ablation) | ImVoxelNet + 3D Copy-Paste · SUN RGB-D | mAP@0.25: 41.80 -> 43.79 | Improved |
| use plausible geometry instead of random insertion (ablation) | ImVoxelNet · SUN RGB-D | mAP@0.25: 37.02 -> 41.80 | Improved |
| remove shadows from the dynamic light rendering (ablation) | 3D Copy-Paste · SUN RGB-D | mAP@0.25: 43.79 -> 41.83 | Hurt |
| vary light type and direction (point, area, spot, sun) instead of estimated light (ablation) | 3D Copy-Paste · SUN RGB-D | mAP@0.25: 40.92 - 42.67 | Mixed |
| follow global scene context when choosing inserted class (ablation) | 3D Copy-Paste · SUN RGB-D | mAP@0.25: 43.79 -> 43.75 | No clear effect |
| select inserted class based on empty floor size, without global context (ablation) | 3D Copy-Paste · SUN RGB-D | mAP@0.25: 43.79 -> 42.50 | Hurt |
| add 3D Copy-Paste augmentation | ImVoxelNet · ScanNet (8 classes) | mAP@0.25: 14.1 -> 16.9 | Improved |
| add 3D Copy-Paste augmentation | Im3D · SUN RGB-D | mAP: 42.13 -> 43.34 | Improved |
| add 3D Copy-Paste augmentation, evaluated at IoU 0.15 | ImVoxelNet · SUN RGB-D | mAP@0.15: 48.45 -> 51.16 | Improved |
| add 3D Copy-Paste augmentation for the bathtub class | ImVoxelNet · SUN RGB-D | AP@0.25: 29.6 -> 40.2 | Improved |
Closest papers by meaning
- MonoPlace3D: Learning 3D-Aware Object Placement for 3D Monocular Detection (CVPR 2025 · similarity 0.89)
- Cubify Anything: Scaling Indoor 3D Object Detection (CVPR 2025 · similarity 0.86)
- MonoSOWA: Scalable Monocular 3D Object Detector Without Human Annotations (ICCV 2025 · similarity 0.86)
- 3D-MOOD: Lifting 2D to 3D for Monocular Open-Set Object Detection (ICCV 2025 · similarity 0.86)
- Rooms from Motion: Un-posed Indoor 3D Object Detection as Localization and Mapping (NeurIPS 2025 · similarity 0.86)
- MonoDiff: Monocular 3D Object Detection and Pose Estimation with Diffusion Models (CVPR 2024 · similarity 0.86)
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.
