Research map / Point clouds / Segmentation and detection
3D Object Detection With Pointformer
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 |
|---|---|---|---|
| Replace PointNet++ with Pointformer in PointRCNN | PointRCNN · KITTI val split, Car (IoU=0.7) | AP: 88.88 -> 90.05; 78.63 -> 79.65; 77.38 -> 78.89 | Helped |
| Replace baseline backbone with Pointformer | CBGS · nuScenes | mAP: 52.8 -> 53.6 | Helped |
| Replace VoteNet backbone with Pointformer | VoteNet · SUN RGB-D validation, AP@0.25 | mAP: 59.1 -> 61.1 | Helped |
| Replace VoteNet backbone with Pointformer | VoteNet · ScanNetV2 validation, AP@0.25 | mAP: 62.9 -> 64.1 | Helped |
| Add Local Transformer (LT) (ablation) | PointRCNN · KITTI val split, Car easy (IoU=0.7) | AP: 88.88 -> 89.46 | Helped |
| Add coordinate refinement (CoRe) to LT (ablation) | PointRCNN + LT · KITTI val split, Car moderate (IoU=0.7) | AP: 78.91 -> 79.24 | Helped |
| Add Global Transformer (GT) to LT (ablation) | PointRCNN + LT · KITTI val split, Car hard (IoU=0.7) | AP: 77.65 -> 78.52 | Helped |
| Add Local-Global Transformer (LGT) to LT and GT (ablation) | PointRCNN + LT + GT · KITTI val split, Car moderate (IoU=0.7) | AP: 79.22 -> 79.34 | Helped |
| Add coordinate refinement (CoRe) to LT, GT, and LGT (ablation) | PointRCNN + LT + GT + LGT · KITTI val split, Car moderate (IoU=0.7) | AP: 79.34 -> 79.65 | Helped |
| Remove positional encoding (ablation) | Pointformer · KITTI val split, Car moderate (IoU=0.7) | AP: 79.65 -> 75.67 | Hurt |
| Reduce the number of Transformer layers (ablation) | Pointformer · KITTI val split, Car moderate (IoU=0.7) | AP: 79.65 -> 79.01 | Hurt |
| Replace standard attention with Linformer (ablation) | Pointformer · KITTI val split, Car moderate (IoU=0.7) | AP: 79.65 -> 79.63 | No clear effect |
Closest papers by meaning
- SEFormer: Structure Embedding Transformer for 3D Object Detection (AAAI 2023 · similarity 0.90)
- Point Transformer V3: Simpler Faster Stronger (CVPR 2024 · similarity 0.90)
- PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · similarity 0.90)
- PTNET: A PROPOSAL-CENTRIC TRANSFORMER NET- WORK FOR 3D OBJECT DETECTION (ICLR 2026 · similarity 0.89)
- Uni3DETR: Unified 3D Detection Transformer (NeurIPS 2023 · similarity 0.89)
- PTT: Point-Trajectory Transformer for Efficient Temporal 3D Object Detection (CVPR 2024 · similarity 0.87)
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.