Research map / Architectures and efficiency / Attention and transformers
MB-TaylorFormer: Multi-Branch Efficient Transformer Expanded by Taylor Formula for Image Dehazing
Yuwei Qiu, Kaihao Zhang, Chenxi Wang, Wenhan Luo, Hongdong Li, Zhi Jin
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 multi-scale attention refinement (MSAR) module (ablation) | MB-TaylorFormer-B · SOTS-Indoor | PSNR: 38.74 -> 40.71 | Improved |
| use deformable depthwise separable serial convs (DSDCN-S) in multi-scale patch embedding instead of single-branch conv (ablation) | MB-TaylorFormer-B · SOTS-Indoor | PSNR: 38.27 -> 40.71 | Improved |
| use deformable convs instead of standard convs in serial patch embedding (ablation) | MB-TaylorFormer-B · SOTS-Indoor | PSNR: 39.04 -> 40.71 | Improved |
| enlarge gating tensor G to C channels (ablation) | MSAR · SOTS-Indoor | PSNR: 40.71 -> 40.88, params 2.677M -> 4.288M | No clear effect |
| use Taylor expanded self-attention instead of other linear attentions (ablation) | MB-TaylorFormer-B · SOTS-Indoor | PSNR: 38.57 (MDTA) -> 40.71 | Improved |
| second-order instead of first-order Taylor expansion in Swin attention (ablation) | Swin · SOTS-Indoor | PSNR: 36.37 -> 36.50 | No clear effect |
| Taylor approximation versus exact softmax in Swin (ablation) | Swin · SOTS-Indoor | PSNR: 36.59 -> 36.37 | Hurt |
| MB-TaylorFormer-B versus SGID-PFF on SOTS-Indoor | MB-TaylorFormer-B · SOTS-Indoor | PSNR: 38.52 -> 40.71 | Improved |
Closest papers by meaning
- Xformer: Hybrid X-Shaped Transformer for Image Denoising (ICLR 2024 · similarity 0.86)
- Prior-guided Hierarchical Harmonization Network for Efficient Image Dehazing (AAAI 2025 · similarity 0.86)
- Scaling White-Box Transformers for Vision (NeurIPS 2024 · similarity 0.85)
- Rethinking Multi-Scale Representations in Deep Deraining Transformer (AAAI 2024 · similarity 0.85)
- An Image is Worth More Than 16x16 Patches: Exploring Transformers on Individual Pixels (ICLR 2025 · similarity 0.85)
- SLAB: Efficient Transformers with Simplified Linear Attention and Progressive Re-parameterized Batch Normalization (ICML 2024 · similarity 0.85)
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.
