Research map / Detection and segmentation / CNN features
Dual Attention Network for Scene Segmentation
Check 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 both position and channel attention modules (ablation) | Dilated FCN with ResNet-50 · Cityscapes val | Mean IoU%: 70.03 -> 76.34 | Helped |
| Add position attention module (ablation) | Dilated FCN with ResNet-50 · Cityscapes val | Mean IoU%: 70.03 -> 75.74 | Helped |
| Add channel attention module (ablation) | Dilated FCN with ResNet-50 · Cityscapes val | Mean IoU%: 70.03 -> 74.28 | Helped |
| Use both attention modules rather than either alone (ablation) | DANet with ResNet-50 · Cityscapes val | Mean IoU%: 75.74 -> 76.34; 74.28 -> 76.34 | Helped |
| Add both position and channel attention modules (ablation) | Dilated FCN with ResNet-101 · Cityscapes val | Mean IoU%: 72.54 -> 77.57 | Helped |
| Add random-scale data augmentation (ablation) | DANet-101 · Cityscapes val | Mean IoU%: 77.57 -> 78.83 | Helped |
| Add Multi-Grid (ablation) | DANet-101 with random-scale augmentation · Cityscapes val | Mean IoU%: 78.83 -> 79.94 | Helped |
| Add multi-scale inference (ablation) | DANet-101 with random-scale augmentation and Multi-Grid · Cityscapes val | Mean IoU%: 79.94 -> 81.50 | Helped |
| Add both attention modules (ablation) | Dilated FCN with ResNet-50 · PASCAL VOC 2012 val | Mean IoU%: 75.7 -> 79.0 | Helped |
| Add attention modules (ablation) | Dilated FCN with ResNet-50 · PASCAL Context test | Mean IoU%: 44.3 -> 50.1 | Helped |
| Add attention modules (ablation) | Dilated FCN with ResNet-50 · COCO Stuff test | Mean IoU%: 31.9 -> 37.2 | Helped |
Closest papers by meaning
- Fully Attentional Network for Semantic Segmentation (AAAI 2022 · similarity 0.89)
- Cars Can't Fly Up in the Sky: Improving Urban-Scene Segmentation via Height-Driven Attention Networks (CVPR 2020 · similarity 0.89)
- AttaNet: Attention-Augmented Network for Fast and Accurate Scene Parsing (AAAI 2021 · similarity 0.87)
- Squeeze-and-Attention Networks for Semantic Segmentation (CVPR 2020 · similarity 0.87)
- Channelized Axial Attention – considering Channel Relation within Spatial Attention for Semantic Segmentation (AAAI 2022 · similarity 0.86)
- Contextrast: Contextual Contrastive Learning for Semantic Segmentation (CVPR 2024 · similarity 0.84)
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.