Research map / Detection and segmentation / CNN features
Cross-Image Pixel Contrasting for Semantic Segmentation
Tianfei Zhou, Wenguan Wang
Read the paperCheck 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 |
|---|---|---|---|
| add intra-image pixel contrast (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 78.1 -> 78.9 | Improved |
| add inter-image pixel contrast (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 78.1 -> 81.0 | Improved |
| add pixel memory to mini-batch contrast (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 79.8 -> 80.5 | Improved |
| add region memory to mini-batch contrast (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 79.8 -> 80.2 | Improved |
| use both pixel and region memory (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 80.5 -> 81.0 | Improved |
| use semi-hard instead of random pos/neg sampling (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 79.3 -> 80.1 | Improved |
| use segmentation-aware hard anchors instead of random (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 80.1 -> 81.0 | Improved |
| use semi-hard instead of hardest sampling (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 80.5 -> 81.0 | Improved |
| use contrastive loss instead of Lovasz loss (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 80.3 -> 81.0 | Improved |
| add AAF loss on top of contrastive loss (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 81.0 -> 81.0 | No clear effect |
| add Lovasz loss on top of contrastive loss (ablation) | HRNetV2-W48 · Cityscapes val (40K iters) | mIoU: 81.0 -> 81.5 | Improved |
| add pixel contrast to OCR | OCR (HRNetV2-W48) · Cityscapes test (train+val) | mIoU: 82.4 -> 83.2 | Improved |
Closest papers by meaning
- Exploring Cross-Image Pixel Contrast for Semantic Segmentation (ICCV 2021 · similarity 0.99)
- Region-Aware Contrastive Learning for Semantic Segmentation (ICCV 2021 · similarity 0.91)
- Looking Beyond Single Images for Contrastive Semantic Segmentation Learning (NeurIPS 2021 · similarity 0.91)
- Contrastive Learning for Label Efficient Semantic Segmentation (ICCV 2021 · similarity 0.90)
- SePiCo: Semantic-Guided Pixel Contrast for Domain Adaptive Semantic Segmentation (TPAMI 2023 · similarity 0.90)
- Hierarchy-Agnostic Unsupervised Segmentation: Parsing Semantic Image Structure (NeurIPS 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.
