Research map / Point clouds / Segmentation and detection
Bridged Transformer for Vision and Point Cloud 3D Object Detection
Yikai Wang, TengQi Ye, Lele Cao, Wenbing Huang, Fuchun Sun, Fengxiang He, Dacheng Tao
Check 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 |
|---|---|---|---|
| Bridge image and point features with conditional queries, shared positional embeddings, and point-to-patch projection (ablation) | BrT · SUN RGB-D | mAP@0.25: 62.6 -> 65.4 | Improved |
| Bridge image and point features with conditional queries, shared positional embeddings, and point-to-patch projection (ablation) | BrT · SUN RGB-D | mAP@0.5: 44.3 -> 48.1 | Improved |
| Use conditional object queries together with shared positional embeddings (ablation) | BrT · SUN RGB-D | mAP@0.25: 62.6 -> 64.0 | Improved |
| Use conditional object queries together with shared positional embeddings (ablation) | BrT · SUN RGB-D | mAP@0.5: 44.3 -> 46.3 | Improved |
| Add point-to-patch projection (ablation) | BrT · SUN RGB-D | mAP@0.25: 62.6 -> 64.2 | Improved |
| Add point-to-patch projection (ablation) | BrT · SUN RGB-D | mAP@0.5: 44.3 -> 46.6 | Improved |
| Use ImageNet-1k ViT-S pre-training rather than training from scratch (ablation) | BrT · SUN RGB-D | mAP@0.25: 62.8 -> 65.4 | Improved |
| Use COCO-finetuned YOLOS-S initialization instead of ImageNet-1k ViT-S initialization (ablation) | BrT · SUN RGB-D | mAP@0.25: 65.4 -> 63.4 | Hurt |
| Directly append 3-dimensional RGB values to Group-Free point features | Group-Free · SUN RGB-D | mAP@0.25: 63.0 -> 61.7 | Hurt |
| Add Faster R-CNN region features to Group-Free seed points | Group-Free · ScanNetV2 | mAP@0.25: 69.1 -> 68.7 | No clear effect |
| Use ViT-B/16 rather than ViT-S/16 | BrT · ScanNetV2 | mAP@0.25: 69.7 -> 71.3 | Improved |
| Allow direct attention connections between point tokens and image patch tokens (ablation) | BrT · SUN RGB-D | mAP@0.25: 65.4 -> 63.9 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- imVoteNet — ImVoteNet: Boosting 3D Object Detection in Point Clouds With Image Votes (CVPR 2020) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- An End-to-End Transformer Model for 3D Object Detection (ICCV 2021 · similarity 0.90)
- 3D Object Detection With Pointformer (CVPR 2021 · similarity 0.88)
- Improving 3D Object Detection With Channel-Wise Transformer (ICCV 2021 · similarity 0.88)
- PVT-SSD: Single-Stage 3D Object Detector With Point-Voxel Transformer (CVPR 2023 · similarity 0.88)
- ImVoteNet: Boosting 3D Object Detection in Point Clouds With Image Votes (CVPR 2020 · similarity 0.87)
- MSV-PCT: Multi-Sparse-View Enhanced Transformer Framework for Salient Object Detection in Point Clouds (AAAI 2025 · 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.
