Research map / Detection and segmentation / CNN features
Revisiting Superpixels for Active Learning in Semantic Segmentation With Realistic Annotation Costs
Lile Cai, Xun Xu, Jun Hao Liew, Chuan Sheng Foo
Check 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 superpixels with dominant labeling instead of rectangles with precise polygon labeling | Region-based active learning · Cityscapes | annotation cost to obtain 95% accuracy (% of total clicks): 15.70% -> 9.81% | Improved |
| Add class-balanced selection to superpixel dominant-label sampling | Superpixel-based active learning · Cityscapes | annotation cost to obtain 95% accuracy (% of total clicks): 9.81% -> 7.85% | Improved |
| Measure annotation cost in clicks rather than labeled-pixel percentage | Superpixel-based versus rectangle-based active learning · Cityscapes | mIoU | Mixed |
| Use dominant labeling rather than precise polygon labeling (ablation) | Rectangular-region active learning · Cityscapes | mIoU | Hurt |
| Use class-balanced selection rather than uncertainty selection (ablation) | Superpixel-based active learning · PASCAL VOC 2012 | mIoU | No clear effect |
| Use smaller regions rather than larger regions (ablation) | Superpixel and rectangle region-based active learning · Cityscapes | mIoU | Improved |
| Increase superpixel region size beyond 128 × 128 (ablation) | Superpixel-based active learning · Cityscapes | mIoU | Hurt |
| Add class-balanced sampling (ablation) | Superpixel-based active learning · Cityscapes | class-wise IoU | Mixed |
| Use uncertainty selection instead of random selection (ablation) | Rectangle+polygon active learning · PASCAL VOC 2012 | mIoU | Hurt |
Closest papers by meaning
- Efficient Active Domain Adaptation for Semantic Segmentation by Selecting Information-rich Superpixels (ECCV 2024 · similarity 0.90)
- ViewAL: Active Learning With Viewpoint Entropy for Semantic Segmentation (CVPR 2020 · similarity 0.90)
- Active Learning for Semantic Segmentation with Multi-class Label Query (NeurIPS 2023 · similarity 0.89)
- Pixels to Tokens: Token Space Efficient Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2026 · similarity 0.89)
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2025 · similarity 0.88)
- Active Label Correction for Semantic Segmentation with Foundation Models (ICML 2024 · similarity 0.88)
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.
