Research map / Detection and segmentation / Weakly supervised detection and segmentation
Weakly Supervised 3D Semantic Segmentation Using Cross-Image Consensus and Inter-Voxel Affinity Relations
Xiaoyu Zhu, Jeffrey Chen, Xiangrui Zeng, Junwei Liang, Chengqi Li, Sinuo Liu, Sima Behpour, Min Xu
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 cross-image co-occurrence learning (CO) (ablation) | Grad-CAM · Simulated cryo-ET (SNR 0.03) | mIoU: 14.8 -> 17.9 | Improved |
| add cross-image co-occurrence learning (CO) (ablation) | Grad-CAM · Simulated cryo-ET (SNR 0.05) | mIoU: 9.7 -> 18.3 | Improved |
| add cross-image co-occurrence learning (CO) (ablation) | Grad-CAM · Real cryo-ET (Poly-GA) | mIoU: 19.0 -> 20.5 | Improved |
| add inter-voxel affinity learning (IVA) (ablation) | Grad-CAM + CO · Simulated cryo-ET (SNR 0.03) | mIoU: 17.9 -> 20.6 | Improved |
| add inter-voxel affinity learning (IVA) (ablation) | Grad-CAM + CO · Simulated cryo-ET (SNR 0.05) | mIoU: 18.3 -> 24.4 | Improved |
| add inter-voxel affinity learning (IVA) (ablation) | Grad-CAM + CO · Real cryo-ET (Poly-GA) | mIoU: 20.5 -> 36.1 | Improved |
| add 3D-CRF post-processing (ablation) | CIVA-Net · Simulated cryo-ET (SNR 0.03) | mIoU: 20.6 -> 39.9 | Improved |
| add 3D-CRF post-processing (ablation) | CIVA-Net · Simulated cryo-ET (SNR 0.05) | mIoU: 24.4 -> 38.8 | Improved |
| add 3D-CRF post-processing (ablation) | CIVA-Net · Real cryo-ET (Poly-GA) | mIoU: 36.1 -> 62.7 | Improved |
| train with voxel affinity pairs instead of cross-entropy on pseudo labels (ablation) | VoxResNet segmentation · Simulated cryo-ET (SNR 0.03) | mIoU: 18.1 -> 20.6 | Improved |
| train with voxel affinity pairs instead of cross-entropy on pseudo labels (ablation) | VoxResNet segmentation · Simulated cryo-ET (SNR 0.05) | mIoU: 20.3 -> 24.4 | Improved |
| train with voxel affinity pairs instead of cross-entropy on pseudo labels (ablation) | VoxResNet segmentation · Real cryo-ET (Poly-GA) | mIoU: 22.7 -> 36.1 | Improved |
Closest papers by meaning
- CIAN: Cross-Image Affinity Net for Weakly Supervised Semantic Segmentation (AAAI 2020 · similarity 0.88)
- Graph-Constrained Contrastive Regularization for Semi-Weakly Volumetric Segmentation (ECCV 2022 · similarity 0.88)
- Rewis3d: Reconstruction Improves Weakly-Supervised Semantic Segmentation (CVPR 2026 · similarity 0.87)
- Multi-Modality Affinity Inference for Weakly Supervised 3D Semantic Segmentation (AAAI 2024 · similarity 0.87)
- 3D Instance Segmentation via Multi-Task Metric Learning (ICCV 2019 · similarity 0.86)
- ToNNO: Tomographic Reconstruction of a Neural Network's Output for Weakly Supervised Segmentation of 3D Medical Images (CVPR 2024 · similarity 0.85)
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.
