Research map / Detection and segmentation / CNN features
Coordinate Attention for Efficient Mobile Network Design
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 coordinate attention instead of SE | MobileNetV2 · ImageNet | Top-1 accuracy: 73.5_{+1.2} -> 74.3_{+2.0} | Helped |
| Use both horizontal and vertical attention rather than only one direction (ablation) | MobileNetV2 coordinate attention · ImageNet | Top-1 accuracy: 73.5_{+1.2} -> 74.3_{+2.0} | Helped |
| Replace coordinate attention with horizontal-only attention (ablation) | MobileNetV2 coordinate attention · ImageNet | Top-1 accuracy: 74.3_{+2.0} -> 73.5_{+1.2} | Hurt |
| Replace coordinate attention with vertical-only attention (ablation) | MobileNetV2 coordinate attention · ImageNet | Top-1 accuracy: 74.3_{+2.0} -> 73.5_{+1.2} | Hurt |
| Reduce the coordinate-attention reduction ratio from 32 to 16 (ablation) | MobileNetV2 + CA · ImageNet | Top-1 accuracy: 74.3_{+2.0} -> 74.7_{+2.4} | Helped |
| Add coordinate attention instead of SE | MobileNeXt · ImageNet | Top-1 accuracy: 74.7_{+0.7} -> 75.2_{+1.2} | Helped |
| Replace SE with coordinate attention | EfficientNet-b0 · ImageNet | Top-1 accuracy: 76.3 -> 76.9 | Helped |
| Add coordinate attention | MobileNetV2 · COCO 2017 validation, SSDLite320 | AP: 22.3 -> 24.5 | Helped |
| Add SE or CBAM | MobileNetV2 · Pascal VOC 2007 test, SSDLite320 | mAP: 71.7 -> 71.7 | No clear effect |
| Add coordinate attention | MobileNetV2 · Pascal VOC 2007 test, SSDLite320 | mAP: 71.7 -> 73.1 | Helped |
| Add coordinate attention instead of SE | MobileNetV2 · Pascal VOC 2012 validation, output stride 16 | mIoU: 71.69 -> 73.32 | Helped |
| Add coordinate attention instead of SE | MobileNetV2 · Cityscapes validation, output stride 8 | mIoU: 72.2 -> 74.0 | Helped |
Closest papers by meaning
- MCA: Moment Channel Attention Networks (AAAI 2024 · similarity 0.85)
- Mobile Attention: Mobile-Friendly Linear-Attention for Vision Transformers (ICML 2024 · similarity 0.85)
- Parallel Sequence Modeling via Generalized Spatial Propagation Network (CVPR 2025 · similarity 0.82)
- Faster Neighborhood Attention: Reducing the O(n^2) Cost of Self Attention at the Threadblock Level (NeurIPS 2024 · similarity 0.82)
- IFORMER: INTEGRATING CONVNET AND TRANSFORMER FOR MOBILE APPLICATION (ICLR 2025 · similarity 0.82)
- LSNet: See Large, Focus Small (CVPR 2025 · similarity 0.82)
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.