Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Topology-Aware Uncertainty for Image Segmentation
Saumya Gupta, Yikai Zhang, Xiaoling Hu, Prateek Prasanna, Chao Chen
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 |
|---|---|---|---|
| replace classic DMT with Probabilistic DMT (ablation) | structure-wise uncertainty model · DRIVE | ECE (%): 6.3481 -> 4.1633 | Improved |
| replace classic DMT with Probabilistic DMT (ablation) | structure-wise uncertainty model · DRIVE | clDice: 0.7729 -> 0.7974 | Improved |
| replace MLP with GNN for joint inference (ablation) | structure-wise uncertainty model · DRIVE | ECE (%): 4.8202 -> 4.1633 | Improved |
| replace MLP with GNN for joint inference (ablation) | structure-wise uncertainty model · DRIVE | clDice: 0.7745 -> 0.7974 | Improved |
| use non-zero distance regularizer gamma in perturb-and-walk (ablation) | Probabilistic DMT · DRIVE | ECE (%) | Improved |
| input feature size and crop 16 / 16x16 instead of 32 / 32x32 (ablation) | structure-wise uncertainty model · DRIVE | ECE / clDice | Hurt |
| increase input feature size and crop to 64 / 64x64 (ablation) | structure-wise uncertainty model · DRIVE | ECE / clDice | No clear effect |
| add structure-wise uncertainty module on top of UNet backbone | UNet · DRIVE | clDice: 0.7586 -> 0.7974 | Improved |
| add module on CS2-Net backbone | CS2-Net · PARSE | Dice: 0.7630 -> 0.7720 | No clear effect |
| add module on UNet backbone | UNet · ROADS | Dice: 0.7011 -> 0.7461 | Improved |
| use proposed method instead of Hu et al. for uncertainty | UNet · DRIVE | ECE (%): 8.0883 -> 4.1633 | Improved |
Closest papers by meaning
- Topology-Aware Segmentation Using Discrete Morse Theory (ICLR 2021 · similarity 0.91)
- Learning Probabilistic Topological Representations Using Discrete Morse Theory (ICLR 2023 · similarity 0.90)
- Topograph: An Efficient Graph-Based Framework for Strictly Topology Preserving Image Segmentation (ICLR 2025 · similarity 0.87)
- Joint Topology-Preserving and Feature-Refinement Network for Curvilinear Structure Segmentation (ICCV 2021 · similarity 0.87)
- Uncertainty-aware Fine-tuning of Segmentation Foundation Models (NeurIPS 2024 · similarity 0.85)
- ValUES: A Framework for Systematic Validation of Uncertainty Estimation in Semantic Segmentation (ICLR 2024 · 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.
