Research map / Detection and segmentation / Weakly supervised detection and segmentation
GeoCoBox: Box-supervised 3D Tumor Segmentation via Geometric Co-embedding
Tianzhong Lan, Zhang Yi, Xiuyuan Xu, Min Zhu
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 |
|---|---|---|---|
| use soft margin + Dice loss for AD-CAM instead of SCE + BCE (ablation) | GeoCoBox AD-CAM · LIDC-IDRI | DSC (%): 68.25 -> 76.09 | Improved |
| replace BCE with Dice loss in AD-CAM (keeping SCE) (ablation) | GeoCoBox AD-CAM · LIDC-IDRI | DSC (%): 68.25 -> 73.92 | Improved |
| use GCL refining scale tau=8 rather than 4 (ablation) | GeoCoBox GCL · LIDC-IDRI | DSC (%): 73.21 -> 76.09 | Improved |
| use GCL refining scale tau=8 rather than 16 (ablation) | GeoCoBox GCL · LIDC-IDRI | DSC (%): 74.32 -> 76.09 | Improved |
| use intensity range delta=20HU rather than 10HU or 40HU (ablation) | GeoCoBox GCL · LIDC-IDRI | DSC (%): 72.89 / 75.22 -> 76.09 | Improved |
| use cosine similarity instead of SSIM or MSE (ablation) | GeoCoBox GCL · LIDC-IDRI | DSC (%): 74.10 (SSIM) / 75.60 (MSE) -> 76.09 | Improved |
| add AD-CAM and GCL components to Vanilla baseline (ablation) | GeoCoBox · LIDC-IDRI | DSC and HD95 | Improved |
| replace 3D UNet backbone with 3D UNETR | GeoCoBox · MSD-Pancreas | HD95 | Mixed |
| train on LIDC-IDRI and test on MSD-Lung (cross-dataset) | GeoCoBox · MSD-Lung | DSC (%): 50.99 -> 54.87 | Improved |
| restrict contrastive refinement to center and boundary points | GeoCoBox GCL · parameter count versus GPCS | parameters | Improved |
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.
- BoxInst — BoxInst: High-Performance Instance Segmentation With Box Annotations (CVPR 2021) · better on 4 of 4 shared results
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
- Wavelet-Driven Cross-Domain Consistency for Mixed-Supervised 3D Tumor Segmentation (ECCV 2026 · similarity 0.89)
- ZePT: Zero-Shot Pan-Tumor Segmentation via Query-Disentangling and Self-Prompting (CVPR 2024 · similarity 0.85)
- Neighbor Does Matter: Density-Aware Contrastive Learning for Medical Semi-supervised Segmentation (AAAI 2025 · similarity 0.85)
- Advancing Generalizable Tumor Segmentation with Anomaly-Aware Open-Vocabulary Attention Maps and Frozen Foundation Diffusion Models (CVPR 2025 · similarity 0.84)
- Integrating Deep Metric Learning with Coreset for Active Learning in 3D Segmentation (NeurIPS 2024 · similarity 0.84)
- Weakly Supervised Gland Segmentation with Class Semantic Consistency and Purified Labels Filtration (AAAI 2025 · similarity 0.84)
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.
