Research map / Detection and segmentation / Weakly supervised detection and segmentation
Adaptive Early-Learning Correction for Segmentation From Noisy Annotations
Check 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 adaptive, class-wise label correction with multiscale consistency | UNet trained on noisy SegTHOR annotations · SegTHOR test | mIoU: 63.3$\pm$2.0 -> 71.2$\pm$0.6 | Helped |
| Correct labels for all classes at the same time instead of class-adaptively (ablation) | ADELE on SegTHOR · SegTHOR test | mIoU: 71.2$\pm$0.6 -> 40.7$\pm$2.4 | Hurt |
| Apply ADELE to SEAM-generated annotations | SEAM / ResNet-38 · PASCAL VOC 2012 val | mIoU: 64.5 -> 69.3 | Helped |
| Apply ADELE to ICD-generated annotations | ICD / ResNet-38 · PASCAL VOC 2012 test | mIoU: 65.5 -> 68.9 | Helped |
| Add multiscale input augmentation (ablation) | SegTHOR segmentation baseline · SegTHOR val | mIoU: 58.8 -> 60.7 | Helped |
| Add multiscale consistency regularization (ablation) | SegTHOR segmentation baseline with multiscale inputs · SegTHOR val | mIoU: 60.7 -> 62.5 | Helped |
| Add adaptive label correction (ablation) | SegTHOR single-scale baseline · SegTHOR val | mIoU: 58.8 -> 65.2 | Helped |
| Add adaptive label correction and multiscale consistency rather than no correction with multiscale consistency (ablation) | SegTHOR · SegTHOR val | mIoU: 62.5 -> 72.2 | Helped |
| Add multiscale consistency regularization (ablation) | PASCAL VOC 2012 segmentation baseline with multiscale inputs · PASCAL VOC 2012 val | mIoU: 65.5 -> 66.7 | Helped |
| Add adaptive label correction (ablation) | PASCAL VOC 2012 single-scale baseline · PASCAL VOC 2012 val | mIoU: 64.5 -> 65.6 | Helped |
| Add adaptive label correction and multiscale consistency rather than no correction with multiscale consistency (ablation) | PASCAL VOC 2012 · PASCAL VOC 2012 val | mIoU: 66.7 -> 69.3 | Helped |
| Use per-iteration rather than epoch-end model outputs for label correction (ablation) | ADELE on SegTHOR · SegTHOR test | mIoU: 70.2 -> 71.2 | Helped |
Closest papers by meaning
- Latent Noise Segmentation: How Neural Noise Leads to the Emergence of Segmentation and Grouping (ICML 2024 · similarity 0.86)
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2025 · similarity 0.86)
- Uncertainty-aware Fine-tuning of Segmentation Foundation Models (NeurIPS 2024 · similarity 0.86)
- Pre-Trained Vision-Language Models as Noisy Partial Annotators (AAAI 2025 · similarity 0.86)
- Test-Time Adaptation of Vision-Language Models for Open-Vocabulary Semantic Segmentation (NeurIPS 2025 · similarity 0.85)
- Towards Fairness under Label Bias in Image Segmentation: Impact, Measurement and Mitigation (NeurIPS 2026 · 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.