Research map / Detection and segmentation / Weakly supervised detection and segmentation
Scribble-Supervised Semantic Segmentation by Uncertainty Reduction on Neural Representation and Self-Supervision on Neural Eigenspace
Zhiyi Pan, Peng Jiang, Yunhai Wang, Changhe Tu, Anthony G. Cohn
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 entropy minimization (ablation) | ResNet50 baseline · ScribbleSup val | mIoU (%): 66.8 -> 69.4 | Improved |
| add random walk (ablation) | ResNet50 baseline · ScribbleSup val | mIoU (%): 66.8 -> 69.3 | Improved |
| add boundary exclusion to entropy minimization (ablation) | ResNet50 · ScribbleSup val | mIoU (%): 69.4 -> 70.0 | Improved |
| add random walk to entropy and boundary exclusion (ablation) | ResNet50 · ScribbleSup val | mIoU (%): 70.0 -> 70.9 | Improved |
| add full uncertainty reduction (ablation) | ResNet50 baseline · ScribbleSup val | mIoU (%): 66.8 -> 70.9 | Improved |
| add self-supervision on neural eigenspace with random transforms (ablation) | Uncertainty-reduction network · ScribbleSup val | mIoU (%): 70.9 -> 73.0 | Improved |
| apply translation self-supervision on f(x)^L instead of eigenspace (ablation) | Uncertainty-reduction network · ScribbleSup val | mIoU (%): 72.6 -> 70.0 | Hurt |
| move self-supervision from f(x)^{L-1} to eigenspace, translation (ablation) | Uncertainty-reduction network · ScribbleSup val | mIoU (%): 71.9 -> 72.6 | Improved |
| move self-supervision from f(x)^{L-1} to eigenspace, flip (ablation) | Uncertainty-reduction network · ScribbleSup val | mIoU (%): 72.5 -> 72.9 | Improved |
| use random rather than flip transform for eigenspace self-supervision (ablation) | Eigenspace self-supervision · ScribbleSup val | mIoU (%): 72.9 -> 73.0 | Improved |
| add uncertainty reduction and self-supervision under scribble drop 0.5 (ablation) | baseline · scribble-drop | mIoU (%): 63.8 -> 69.5 | Improved |
| add uncertainty reduction and self-supervision under scribble shrink 1 (spots) (ablation) | baseline · scribble-shrink | mIoU (%): 57.0 -> 62.8 | 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.
- ScribbleSup — ScribbleSup: Scribble-Supervised Convolutional Networks for Semantic Segmentation (CVPR 2016) · better on 1 of 1 shared result
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
- Scribbles for All: Benchmarking Scribble Supervised Segmentation Across Datasets (NeurIPS 2024 · similarity 0.92)
- Scribble Hides Class: Promoting Scribble-Based Weakly-Supervised Semantic Segmentation with Its Class Label (AAAI 2024 · similarity 0.91)
- Scribble-Supervised Semantic Segmentation with Prototype-based Feature Augmentation (ICML 2024 · similarity 0.90)
- Uncertainty-aware Fine-tuning of Segmentation Foundation Models (NeurIPS 2024 · similarity 0.86)
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2025 · similarity 0.86)
- HSRDiff: A Hierarchical Self-Regulation Diffusion Model for Stochastic Semantic Segmentation (AAAI 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.
