Research map / Detection and segmentation / Open-set and few-shot detection
Better than Average: Spatially-Aware Aggregation of Segmentation Uncertainty Improves Downstream Performance
Vanessa Emanuela Guarino, Claudia Winklmayr, Jannik Franzen, Josef Lorenz Rumberger, Manuel Pfeuffer, Sonja Greven, Klaus Maier-Hein, Dagmar Kainmueller, Christoph Karg, Carsten T. Lüth
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 prediction-based aggregation (BCA/ICA) instead of global average | uncertainty aggregation · OoD detection, 10 datasets | AUROC | Improved |
| use meta-aggregator GMM-All instead of global average | uncertainty aggregation · OoD detection, 10 datasets | AUROC | Improved |
| use global average (AVG) on LIDC-Mal | uncertainty aggregation · LIDC-Mal OoD | AUROC | Improved |
| use GMM-based meta-aggregators on Lizard (LIZ-IG/SG) | GMM-All/Spa/Int · LIZ-IG, LIZ-SG OoD | AUROC | Hurt |
| use above-threshold average (ATA) for OoD detection | uncertainty aggregation · OoD detection, 10 datasets | AUROC | Hurt |
| use ATA for failure detection | uncertainty aggregation · WORM-Nem, ARC-Nuc FD | E-AURC | Improved |
| use QFR (foreground-ratio quantile) instead of BCA/ICA for failure detection | uncertainty aggregation · Failure detection, 10 datasets | E-AURC | Improved |
| add spatial features (GMM-All) to intensity-only features (GMM-Int) (ablation) | GMM meta-aggregator · OoD detection and FD | AUROC / E-AURC | Improved |
| use GMM-All with all aggregators instead of single or leave-one-out features (ablation) | GMM-All · CAR-CS, LIZ-IG, WEED-Hand, WORM-Nem | AUROC | Mixed |
| use global average (AVG) for failure detection | uncertainty aggregation · Failure detection, 10 datasets | E-AURC | Hurt |
Closest papers by meaning
- ValUES: A Framework for Systematic Validation of Uncertainty Estimation in Semantic Segmentation (ICLR 2024 · similarity 0.89)
- Uncertainty-aware Fine-tuning of Segmentation Foundation Models (NeurIPS 2024 · similarity 0.88)
- Arctique: An artificial histopathological dataset unifying realism and controllability for uncertainty quantification (NeurIPS 2024 · similarity 0.86)
- Predictive Uncertainty Quantification for Bird's Eye View Segmentation: A Benchmark and Novel Loss Function (ICLR 2025 · similarity 0.86)
- UncertainSAM: Fast and Efficient Uncertainty Quantification of the Segment Anything Model (ICML 2025 · similarity 0.86)
- UNICL-SAM: Uncertainty-Driven In-Context Segmentation with Part Prototype Discovery (CVPR 2025 · 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.
