Research map / Detection and segmentation / Weakly supervised detection and segmentation
Active Learning for Semantic Segmentation with Multi-class Label Query
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 multi-class labels instead of dominant-class labels | region-based active learning for semantic segmentation · Cityscapes and PASCAL VOC 2012 | mIoU | Helped |
| use PixBal sampling instead of other tested acquisition functions | multi-class labeling active learning · Cityscapes | mIoU | Helped |
| use PixBal sampling instead of BvSB | multi-class labeling active learning · PASCAL VOC 2012 | mIoU | Mixed |
| add label expansion to the full method (ablation) | MulActSeg · Cityscapes, validation, five-round average | mIoU (%): 73.2 -> 72.0 | Helped |
| add intra-region label localization to the method without pseudo labeling (ablation) | MulActSeg · Cityscapes, validation, five-round average | mIoU (%): 72.0 -> 70.4 | Helped |
| add prototypical pixel loss (ablation) | MulActSeg stage 1 · Cityscapes, validation, five-round average | mIoU (%): 70.4 -> 69.2 | Helped |
| add merged positive loss (ablation) | MulActSeg stage 1 · Cityscapes, validation, five-round average | mIoU (%): 70.4 -> 69.4 | Helped |
| use prototype-based localization instead of the most-confident-class baseline (ablation) | MulActSeg pseudo labeling · Cityscapes | mIoU | Helped |
| use multi-class labels rather than dominant-class labels with label expansion (ablation) | pseudo-label expansion · Cityscapes | mIoU | Helped |
| use multi-class labels instead of dominant-class labels | region queries with two or three classes · User study | Time per click (s): 8.02 -> 4.89; 8.60 -> 6.37 | Helped |
| use multi-class labels instead of dominant-class labels | region queries with two or three classes · User study | Accuracy (%): 72.05 -> 87.14; 65.83 -> 71.52 | Helped |
| use SEEDS instead of SLIC superpixels (ablation) | dominant and multi-class labeling · Cityscapes | mIoU | Helped |
Closest papers by meaning
- Active Label Correction for Semantic Segmentation with Foundation Models (ICML 2024 · similarity 0.91)
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2025 · similarity 0.89)
- Multi-Label Prototype Visual Spatial Search for Weakly Supervised Semantic Segmentation (CVPR 2025 · similarity 0.88)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · similarity 0.88)
- Pixels to Tokens: Token Space Efficient Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2026 · similarity 0.87)
- W2P: Switching from Weak Supervision to Partial Supervision for Semantic Segmentation (AAAI 2024 · similarity 0.87)
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.