Research map / Detection and segmentation / CNN features
AttaNet: Attention-Augmented Network for Fast and Accurate Scene Parsing
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 SAM (ablation) | ResNet18 aggregation baseline · Cityscapes val | mIoU: 72.8 -> 77.1; 4.3↑ | Helped |
| add AFM (ablation) | ResNet18 aggregation baseline · Cityscapes val | mIoU: 72.8 -> 77.3; 4.5↑ | Helped |
| add both SAM and AFM (ablation) | ResNet18 aggregation baseline · Cityscapes val | mIoU: 72.8 -> 78.5; 5.7↑ | Helped |
| use vertical rather than horizontal striping in SAM (ablation) | ResNet18 aggregation baseline · Cityscapes val | mIoU: 76.9 -> 77.1 | Helped |
| replace NL with SAM (ablation) | ResNet18 aggregation baseline · Cityscapes val | GFLOPs: 3.357 -> 0.185 | Helped |
| replace RCCA with SAM (ablation) | ResNet18 aggregation baseline · Cityscapes val | GFLOPs: 0.472 -> 0.185 | Helped |
| replace EMA with SAM (ablation) | ResNet18 aggregation baseline · Cityscapes val | GFLOPs: 0.335 -> 0.185 | Helped |
| use AFM instead of simple addition (ablation) | ResNet18 aggregation baseline · Cityscapes val | mIoU: 72.8 -> 77.3 | Helped |
| add AttaNet modules (ablation) | ResNet50 · Cityscapes val | mIoU: 73.4 -> 81.2 | Helped |
| add AttaNet modules (ablation) | ResNet101 · Cityscapes val | mIoU: 74.5 -> 81.0 | Helped |
| add AttaNet modules (ablation) | ShuffleNetV2 · Cityscapes val | mIoU: 67.7 -> 76.0 | Helped |
| add AttaNet modules (ablation) | DF1 · Cityscapes val | mIoU: 70.3 -> 78.0 | Helped |
Closest papers by meaning
- RANet: Region Attention Network for Semantic Segmentation (NeurIPS 2020 · similarity 0.88)
- Squeeze-and-Attention Networks for Semantic Segmentation (CVPR 2020 · similarity 0.88)
- Dual Attention Network for Scene Segmentation (CVPR 2019 · similarity 0.87)
- Hierarchy-Agnostic Unsupervised Segmentation: Parsing Semantic Image Structure (NeurIPS 2024 · similarity 0.86)
- xT: Nested Tokenization for Larger Context in Large Images (ICML 2024 · similarity 0.86)
- ResCLIP: Residual Attention for Training-free Dense Vision-language Inference (CVPR 2025 · 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.