Research map / Detection and segmentation / CNN features
Semantic Diffusion Network for Semantic Segmentation
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 |
|---|---|---|---|
| add SDN neck to FCN | FCN ResNet-50 · ADE20K val | mIoU single-scale: 36.10 -> 38.12 | Helped |
| add SDN neck to SemanticFPN | SemanticFPN ResNet-50 · ADE20K val | mIoU single-scale: 37.49 -> 38.79 | Helped |
| add SDN neck to Segmenter | Segmenter ViT-B · ADE20K val | mIoU single-scale: 48.48 -> 51.05 | Helped |
| add SDN neck to FCN | FCN ResNet-50 · Cityscapes val | mIoU single-scale: 72.64 -> 74.75 | Helped |
| add SDN neck to SemanticFPN | SemanticFPN ResNet-50 · Cityscapes val | mIoU single-scale: 74.10 -> 75.97 | Helped |
| add SDN neck to Segmenter | Segmenter ViT-B · Cityscapes val | mIoU single-scale: 77.97 -> 79.42 | Helped |
| add SDN to OCRNet | OCRNet HRNet-48 · Cityscapes val boundary | F-score 1px: 65.2 -> 69.5 | Helped |
| replace semantic difference convolution with vanilla convolution (ablation) | SDN · Cityscapes val boundary | F-score 1px: 69.5 -> 65.2 | Hurt |
| replace semantic difference convolution with CDC (ablation) | SDN · Cityscapes val boundary | F-score 1px: 69.5 -> 60.1 | Hurt |
| increase SDC kernel size from 3x3 to 5x5 (ablation) | SDN · Cityscapes val boundary | F-score 1px: 69.5 -> 69.6 | No clear effect |
| increase SDC dilation rate to 3 and 5 (ablation) | SDN · Cityscapes val boundary | F-score 1px: 69.5 -> 69.1 (d=3), 68.8 (d=5) | Hurt |
| add SegFix/CRF/InverseForm post-processing on top of SDN (ablation) | SDN · Cityscapes val boundary | F-score 1px: 69.5 -> 69.7 (CRF), 70.1 (SegFix), 70.5 (InverseForm) | Helped |
Closest papers by meaning
- Unleashing the Potential of the Diffusion Model in Few-shot Semantic Segmentation (NeurIPS 2024 · similarity 0.89)
- VidSeg: Training-free Video Semantic Segmentation based on Diffusion Models (CVPR 2025 · similarity 0.89)
- EmerDiff: Emerging Pixel-level Semantic Knowledge in Diffusion Models (ICLR 2024 · similarity 0.89)
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2025 · similarity 0.89)
- $\mathbf{D^{3}S^{2}}$: Diffusion-Guided Dataset Distillation for Semantic Segmentation (NeurIPS 2026 · similarity 0.88)
- An End-to-End Robust Point Cloud Semantic Segmentation Network with Single-Step Conditional Diffusion Models (CVPR 2025 · 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.