Research map / Detection and segmentation / 3D detection and driving
Rotationally Equivariant 3D Object Detection
Check 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 |
|---|---|---|---|
| Add object-level rotation equivariance with rotation equivariance suspension | VoteNet · ScanNetV2 validation | mAP (AP25): boost in mAP by 6.3% | Helped |
| Add object-level rotation equivariance with rotation equivariance suspension | Liu et al. detector · ScanNetV2 validation | mAP (AP25): boost in mAP by 9.0% | Helped |
| Add object-level rotation equivariance with rotation equivariance suspension | VoteNet · SUN RGB-D | mAP (AP25): boost in mAP by 3.1% | Helped |
| Add object-level rotation equivariance with rotation equivariance suspension | PointRCNN · KITTI validation, Pedestrian | AP: 3.02%, 5.15%, 2.90% | Helped |
| Add object-level rotation equivariance with rotation equivariance suspension | PointRCNN · KITTI validation, Cyclist Easy | AP: 87.81 -> 87.33 | Hurt |
| Suspend equivariance at the second-last backbone layer instead of region aggregation (ablation) | EON-VoteNet · ScanNetV2 | mAP@0.25: 56.7 -> 54.5 | Hurt |
| Continue equivariant computation through the whole feature extraction instead of suspending at region aggregation (ablation) | EON-VoteNet · ScanNetV2 | mAP@0.5: 36.5 -> 26.9 | Hurt |
| Use ground-truth orientation labels (ablation) | EON-VoteNet · ScanNetV2 | mAP@0.25: 50.4 -> 64.0 | Helped |
| Use ground-truth foreground segmentation (ablation) | EON-VoteNet · ScanNetV2 | mAP@0.25: 50.4 -> 65.3 | Helped |
| Use ground-truth orientation labels and foreground segmentation (ablation) | EON-VoteNet · ScanNetV2 | mAP@0.25: +19.4% | Helped |
Closest papers by meaning
- R2Det: Exploring Relaxed Rotation Equivariance in 2D Object Detection (ICLR 2025 · similarity 0.90)
- A Functional Approach to Rotation Equivariant Non-Linearities for Tensor Field Networks. (CVPR 2021 · similarity 0.89)
- Relaxed Rotational Equivariance via G-Biases in Vision (AAAI 2025 · similarity 0.89)
- FRED: Towards a Full Rotation-Equivariance in Aerial Image Object Detection (AAAI 2024 · similarity 0.89)
- 3D Equivariant Pose Regression via Direct Wigner-D Harmonics Prediction (NeurIPS 2024 · similarity 0.87)
- A Probabilistic Approach to Learning the Degree of Equivariance in Steerable CNNs (ICML 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.