Research map / Detection and segmentation / CNN features
Region-Aware Contrastive Learning for Semantic Segmentation
Hanzhe Hu, Jinshi Cui, Liwei Wang
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 intra-image region-aware contrastive loss to CE (ablation) | DeepLabV3 ResNet-50 · Cityscapes val (DeepLabV3 R50 ablation) | mIoU: 76.4 -> 77.5 | Improved |
| extend contrast to inter-image with memory bank (ablation) | RegionContrast DeepLabV3 ResNet-50 · Cityscapes val (DeepLabV3 R50 ablation) | mIoU: 77.5 -> 79.6 | Improved |
| build region centers with easy and hard positives instead of direct average (ablation) | RegionContrast · Cityscapes val (DeepLabV3 R50 ablation) | mIoU: 78.2 -> 78.8 | Improved |
| weight easy and hard positives (dynamic sampling) (ablation) | RegionContrast · Cityscapes val (DeepLabV3 R50 ablation) | mIoU: 78.8 -> 79.6 | Improved |
| add memory bank of size 1000 (ablation) | RegionContrast · Cityscapes val (DeepLabV3 R50 ablation) | mIoU: 77.5 -> 78.5 | Improved |
| increase memory bank size from 1000 to 2000 (ablation) | RegionContrast · Cityscapes val (DeepLabV3 R50 ablation) | mIoU: 78.5 -> 79.1 | Improved |
| increase memory bank size from 2000 to full training set 2975 (ablation) | RegionContrast · Cityscapes val (DeepLabV3 R50 ablation) | mIoU: 79.1 -> 79.6 | Improved |
| add RegionContrast to DeepLabV3 ResNet-50 | DeepLabV3 · Cityscapes val | mIoU: 76.4 -> 79.6 | Improved |
| add RegionContrast to DeepLabV3 ResNet-101 | DeepLabV3 · Cityscapes val | mIoU: 79.0 -> 81.3 | Improved |
| add RegionContrast to HRNetV2-W48 | HRNetV2 · Cityscapes val | mIoU: 80.4 -> 81.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DANet — DANet: Divergent Activation for Weakly Supervised Object Localization (ICCV 2019) · better on 2 of 2 shared results
- CCNet — CCNet: Criss-Cross Attention for Semantic Segmentation (ICCV 2019) · 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
- Exploring Cross-Image Pixel Contrast for Semantic Segmentation (ICCV 2021 · similarity 0.91)
- Multi-Scale and Cross-Scale Contrastive Learning for Semantic Segmentation (ECCV 2022 · similarity 0.91)
- Cross-Image Pixel Contrasting for Semantic Segmentation (TPAMI 2024 · similarity 0.91)
- Contrastive Learning for Label Efficient Semantic Segmentation (ICCV 2021 · similarity 0.90)
- Looking Beyond Single Images for Contrastive Semantic Segmentation Learning (NeurIPS 2021 · similarity 0.90)
- Contextrast: Contextual Contrastive Learning for 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.
