Research map / Detection and segmentation / Weakly supervised detection and segmentation
Semi-Supervised Semantic Segmentation With Directional Context-Aware Consistency
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 |
|---|---|---|---|
| Add context-aware directional contrastive consistency with negative sampling and positive filtering | PSPNet ResNet50 · PASCAL VOC 1/8 | mIoU: 65.0 -> 72.4 | Helped |
| Add context-aware directional contrastive consistency | DeepLabv3+ ResNet50 · PASCAL VOC 1/8 | mIoU: 68.3 -> 72.4 | Helped |
| Add context-aware directional contrastive consistency | DeepLabv3+ ResNet50 · Cityscapes 1/8 | mIoU: 66.0 -> 69.7 | Helped |
| Add the method in the full labeled-data setting | DeepLabv3+ ResNet50 · Cityscapes full | mIoU: 77.7 -> 77.5 | Hurt |
| Add contextual augmentation consistency rather than using only low-level transformations (ablation) | Full method, PSPNet ResNet50 · PASCAL VOC 1/8 (ablation) | mIoU: 70.5 -> 72.4 | Helped |
| Add directional alignment to the sampled contrastive loss (ablation) | Contrastive method with negative sampling and positive filtering · PASCAL VOC 1/8 (ablation) | mIoU: 71.2 -> 72.4 | Helped |
| Add pseudo-label negative sampling to avoid false negatives (ablation) | Directional contrastive loss · PASCAL VOC 1/8 (ablation) | mIoU: 64.8 -> 71.6 | Helped |
| Add positive filtering for low-confidence positive pairs (ablation) | Directional contrastive loss with negative sampling · PASCAL VOC 1/8 (ablation) | mIoU: 71.6 -> 72.4 | Helped |
| Add a nonlinear projector (ablation) | Full method · PASCAL VOC 1/8 (ablation) | mIoU: 61.5 -> 72.4 | Helped |
| Replace contrastive alignment with L2 alignment (ablation) | Context-aware consistency · PASCAL VOC 1/8 (ablation) | mIoU: 65.0 -> 64.2 | Hurt |
| Use vanilla contrastive loss without the directional mask or sampling strategies (ablation) | Context-aware consistency · PASCAL VOC 1/8 (ablation) | mIoU: 65.0 -> 56.4 | Hurt |
| Increase the negative sample count from 500 to 19.2k per GPU (ablation) | Full method · PASCAL VOC (negative-count ablation) | mIoU: 70.9 -> 72.4 | Helped |
Closest papers by meaning
- Semi-Supervised Semantic Segmentation With Cross-Consistency Training (CVPR 2020 · similarity 0.93)
- Improving Semi-Supervised Semantic Segmentation with Sliced-Wasserstein Feature Alignment and Uniformity (CVPR 2025 · similarity 0.91)
- ScaleMatch: Multi-scale Consistency Enhancement for Semi-supervised Semantic Segmentation (AAAI 2025 · similarity 0.90)
- RankMatch: Exploring the Better Consistency Regularization for Semi-supervised Semantic Segmentation (CVPR 2024 · similarity 0.89)
- SemiDAViL: Semi-supervised Domain Adaptation with Vision-Language Guidance for Semantic Segmentation (CVPR 2025 · similarity 0.89)
- AllSpark: Reborn Labeled Features from Unlabeled in Transformer for Semi-Supervised Semantic Segmentation (CVPR 2024 · similarity 0.89)
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.