Research map / Detection and segmentation / CNN features
SegNeXt: Rethinking Convolutional Attention Design 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 |
|---|---|---|---|
| Use the full multi-scale convolutional attention module instead of a single large-kernel branch (ablation) | SegNeXt-T · ADE20K | mIoU (SS): 39.5 -> 41.0 | Helped |
| Use the full multi-scale convolutional attention module instead of a single large-kernel branch (ablation) | SegNeXt-S · ADE20K | mIoU (SS): 43.5 -> 44.3 | Helped |
| Add the 7×7 and 11×11 branches alongside the 21×21 branch (ablation) | MSCA · ADE20K | mIoU: 40.0 -> 41.1 | Helped |
| Add the 1×1 channel-mixing convolution (ablation) | MSCA · ADE20K | mIoU: 39.1 -> 41.1 | Helped |
| Add element-wise attention multiplication (ablation) | MSCA · ADE20K | mIoU: 40.5 -> 41.1 | Helped |
| Aggregate encoder features with decoder structure (c), including a lightweight Hamburger module (ablation) | SegNeXt-T · ADE20K | mIoU (SS): 40.3 -> 41.1 | Helped |
| Feed stage 1 features into the decoder (ablation) | SegNeXt-T (c) · ADE20K | mIoU (SS): 41.1 -> 40.7 | Hurt |
| Use Hamburger decoder attention rather than CC attention (ablation) | SegNeXt-B · ADE20K | mIoU (SS): 47.3 -> 48.5 | Helped |
| Use Hamburger decoder attention rather than EMA attention (ablation) | SegNeXt-B · ADE20K | mIoU (SS): 48.0 -> 48.5 | Helped |
Closest papers by meaning
- SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers (NeurIPS 2021 · similarity 0.90)
- Rethinking Semantic Segmentation From a Sequence-to-Sequence Perspective With Transformers (CVPR 2021 · similarity 0.89)
- SegMAN: Omni-scale Context Modeling with State Space Models and Local Attention for Semantic Segmentation (CVPR 2025 · similarity 0.89)
- Rethinking Decoders for Transformer-based Semantic Segmentation: A Compression Perspective (NeurIPS 2024 · similarity 0.88)
- Seg4Diff: Unveiling Open-Vocabulary Semantic Segmentation in Text-to-Image Diffusion Transformers (NeurIPS 2025 · similarity 0.86)
- Seg-ReSearch: Segmentation with Interleaved Reasoning and External Search (ICML 2026 · 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.