Research map / Point clouds / Segmentation and detection
Rewis3d: Reconstruction Improves Weakly-Supervised Semantic Segmentation
Jonas Ernst, Wolfgang Boettcher, Lukas Hoyer, Jan Eric Lenssen, Bernt Schiele
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 |
|---|---|---|---|
| add 3D reconstruction cross-modal consistency (Rewis3d, Recon) | EMA Mean Teacher SegFormer-B4 · Waymo (scribble) | mIoU: 49.4 -> 53.3 | Improved |
| add cross-modal consistency from reconstructed 3D | EMA baseline · Cityscapes (point labels) | mIoU: 50.5 -> 56.5 | Improved |
| add cross-modal consistency from reconstructed 3D | EMA baseline · Cityscapes (coarse labels) | mIoU: 66.5 -> 68.6 | Improved |
| use reconstructed point clouds instead of ground-truth LiDAR | Rewis3d · Waymo (scribble) | mIoU: 51.8 -> 53.3 | Improved |
| add prediction confidence filtering (ablation) | Rewis3d · Waymo (scribble) | mIoU: 51.9 -> 52.7 | Improved |
| add reconstruction confidence filtering (ablation) | Rewis3d · Waymo (scribble) | mIoU: 51.9 -> 52.1 | Improved |
| combine both confidence filters (ablation) | Rewis3d · Waymo (scribble) | mIoU: 51.9 -> 53.3 | Improved |
| view-aware sampling instead of random sampling (ablation) | Rewis3d · Waymo (scribble) | mIoU: 51.9 -> 53.3 | Improved |
| multi-view instead of single-frame reconstruction (ablation) | Rewis3d · Waymo (scribble) | mIoU: 52.1 -> 53.3 | Improved |
| single-frame reconstruction vs no 3D (ablation) | EMA baseline · Waymo (scribble) | mIoU: 49.4 -> 52.1 | Improved |
Closest papers by meaning
- Joint Learning of 2D-3D Weakly Supervised Semantic Segmentation (NeurIPS 2022 · similarity 0.89)
- Reliev3R: Relieving Feed-forward 3D Reconstruction from Multi-View Geometric Annotations (CVPR 2026 · similarity 0.89)
- Geometry-Guided Semantic Reconstruction for 3D Instance Segmentation (NeurIPS 2026 · similarity 0.89)
- DBGroup: Dual-Branch Point Grouping for Weakly Supervised 3D Semantic Instance Segmentation (AAAI 2026 · similarity 0.89)
- Multi-Modality Affinity Inference for Weakly Supervised 3D Semantic Segmentation (AAAI 2024 · similarity 0.88)
- MESC-3D:Mining Effective Semantic Cues for 3D Reconstruction from a Single Image (CVPR 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.
