Research map / Architectures and efficiency / Attention and transformers
Head-Free Lightweight Semantic Segmentation with Linear Transformer
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 |
|---|---|---|---|
| replace Segformer with AFFormer-base | semantic segmentation · ADE20K | mIoU: 37.4 -> 41.8 | Helped |
| replace Segformer with AFFormer-base | semantic segmentation · Cityscapes val | mIoU: 76.2 -> 78.7 | Helped |
| use AFFormer-base instead of lightweight segmentation baselines | semantic segmentation · COCO-stuff | mIoU: 31.5 -> 35.1 | Helped |
| add PD to PL-only structure (ablation) | AFFormer-base · ADE20K | mIoU: 39.2 -> 41.8 | Helped |
| add PL to PD-only structure (ablation) | AFFormer-base · ADE20K | mIoU: 19.5 -> 41.8 | Helped |
| use heterogeneous PD and PL rather than only PD (ablation) | AFFormer-base · ADE20K | mIoU: 27.4 -> 41.8 | Helped |
| use heterogeneous PD and PL rather than only PL (ablation) | AFFormer-base · ADE20K | mIoU: 41.6 -> 41.8 | No clear effect |
| add DHF to DLF-only frequency filtering (ablation) | AFFormer-base · ADE20K | mIoU: 38.7 -> 41.1 | Helped |
| add DHF to DLF-only frequency filtering (ablation) | AFFormer-base · Cityscapes | mIoU: 75.7 -> 77.8 | Helped |
| add FSK to DLF + DHF (ablation) | AFFormer-base · ADE20K | mIoU: 41.1 -> 41.8 | Helped |
| add FSK to DLF + DHF (ablation) | AFFormer-base · Cityscapes | mIoU: 77.8 -> 78.7 | Helped |
| add DLF to DHF + FSK (ablation) | AFFormer-base · ADE20K | mIoU: 41.2 -> 41.8 | Helped |
Closest papers by meaning
- Rethinking Decoders for Transformer-based Semantic Segmentation: A Compression Perspective (NeurIPS 2024 · similarity 0.89)
- Lightweight Frequency Masker for Cross-Domain Few-Shot Semantic Segmentation (NeurIPS 2024 · similarity 0.87)
- Frequency-Adaptive Dilated Convolution for Semantic Segmentation (CVPR 2024 · similarity 0.87)
- MaSS13K: A Matting-level Semantic Segmentation Benchmark (CVPR 2025 · similarity 0.87)
- When Semantic Segmentation Meets Frequency Aliasing (ICLR 2024 · similarity 0.86)
- PEM: Prototype-based Efficient MaskFormer for Image Segmentation (CVPR 2024 · similarity 0.86)
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.