Research map / Detection and segmentation / Open-set and few-shot detection
RankSEG: A Consistent Ranking-based Framework for Segmentation
Ben Dai, Chunlin Li
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 RankDice instead of thresholding at 0.5 | segmentation with true probabilities · Simulation Example 1, step(0.3), 64x64 | Dice: 0.038 -> 0.517 | Improved |
| use RankDice instead of thresholding at 0.5 | segmentation with true probabilities · Simulation Example 1, step(0.1), 64x64 | Dice: 0.083 -> 0.279 | Improved |
| use RankDice instead of thresholding at 0.5 | segmentation with true probabilities · Simulation Example 1, step(0.5), 64x64 | Dice: 0.707 -> 0.707 | No clear effect |
| use RankDice instead of thresholding at 0.5 | segmentation with true probabilities · Simulation Example 1, exp(1.01), 28x28 | Dice: 0.870 -> 0.870 | No clear effect |
| use RankDice instead of thresholding at 0.5 | segmentation with true probabilities · Simulation Example 1, exp(1.01), 256x256 | Dice: 0.286 -> 0.450 | Improved |
| use RankDice instead of thresholding at 0.5 | segmentation with true probabilities · Simulation Example 1, exp(1.05), 64x64 | Dice: 0.296 -> 0.446 | Improved |
| use RankDice instead of thresholding at 0.5 | segmentation with true probabilities · Simulation Example 1, linear(4.00), 64x64 | Dice: 0.580 -> 0.646 | Improved |
| use RankDice instead of the best fixed threshold | segmentation with heterogeneous probabilities · Simulation Example 2 | Dice: 0.560 -> 0.601 | Improved |
| tune the fixed threshold from 0.1 to 0.2 | thresholding framework · Simulation Example 2 | Dice: 0.481 -> 0.560 | Improved |
| raise the fixed threshold from 0.5 to 0.9 | thresholding framework · Simulation Example 2 | Dice: 0.560 -> 0.230 | Hurt |
| faster probability decay (exponential/linear) | RankDice · Simulation Example 1 | Dice gain over thresholding | Improved |
| use RankDice instead of thresholding and argmax | DeepLab, PSPNet, FCN8 on Cityscapes, VOC, Kvasir-SEG · real datasets | mDice and mIoU | Improved |
Closest papers by meaning
- RankSEG-RMA: An Efficient Segmentation Algorithm via Reciprocal Moment Approximation (NeurIPS 2025 · similarity 0.93)
- On the Relaxation of Conditional Independence Assumption for Image Segmentation (NeurIPS 2026 · similarity 0.88)
- Uncertainty-aware Fine-tuning of Segmentation Foundation Models (NeurIPS 2024 · similarity 0.86)
- Learning Hierarchical Image Segmentation For Recognition and By Recognition (ICLR 2024 · similarity 0.85)
- Improving Semi-Supervised Semantic Segmentation with Sliced-Wasserstein Feature Alignment and Uniformity (CVPR 2025 · similarity 0.85)
- Hierarchy-Agnostic Unsupervised Segmentation: Parsing Semantic Image Structure (NeurIPS 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.
