Research map / Architectures and efficiency / CNN design
Gaussian Context Transformer
Dongsheng Ruan, Daiyin Wang, Yuan Zheng, Nenggan Zheng, Min Zheng
Check 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 parameter-free GCT-B0 channel attention | ResNet50 · ImageNet validation | Top-1 accuracy: 76.15 -> 77.51 | Improved |
| Add learnable GCT-B1 channel attention | ResNet50 · ImageNet validation | Top-1 accuracy: 76.15 -> 77.55 | Improved |
| Use learnable rather than fixed Gaussian standard deviation | ResNet50 + GCT · ImageNet validation | Top-1 accuracy: 77.51 -> 77.55 | Improved |
| Add GCT-B0 channel attention | ResNet101 · ImageNet validation | Top-1 accuracy: 77.37 -> 78.60 | Improved |
| Add GCT-B1 channel attention | ResNet101 · ImageNet validation | Top-1 accuracy: 77.37 -> 78.85 | Improved |
| Add GCT-B0 channel attention | ShufflenetV2 · ImageNet validation | Top-1 accuracy: 69.36 -> 70.79 | Improved |
| Add GCT-B0 channel attention | Faster R-CNN ResNet50 · COCO 2017 validation | APbbox 0.5:0.95: 37.4 -> 38.8 | Improved |
| Add GCT-B1 channel attention | Faster R-CNN ResNet50 · COCO 2017 validation | APbbox 0.5:0.95: 37.4 -> 38.9 | Improved |
| Add GCT-B0 channel attention | Mask R-CNN ResNet50 · COCO 2017 validation | APbbox 0.5:0.95: 38.2 -> 39.3 | Improved |
| Add GCT-B0 channel attention | RetinaNet ResNet50 · COCO 2017 validation | APbbox 0.5:0.95: 36.5 -> 38.1 | Improved |
| Add GCT-B0 channel attention | Mask R-CNN ResNet50 · COCO 2017 validation | APmask 0.5:0.95: 34.7 -> 35.8 | Improved |
| Vary fixed Gaussian standard deviation c from 1 to 6; c=2 gives the best reported score (ablation) | GCT-ResNet50-B0 · ImageNet validation | Top-1 accuracy: 77.15, 77.51, 77.24, 77.27, 77.11, 77.06 | Mixed |
Closest papers by meaning
- Linear Context Transform Block (AAAI 2020 · similarity 0.88)
- GPCA: A Probabilistic Framework for Gaussian Process Embedded Channel Attention (TPAMI 2021 · similarity 0.87)
- G2DeNet: Global Gaussian Distribution Embedding Network and Its Application to Visual Recognition (CVPR 2017 · similarity 0.86)
- Gated Linear Attention Transformers with Hardware-Efficient Training (ICML 2024 · similarity 0.86)
- Contextual Transformer Networks for Visual Recognition (TPAMI 2022 · similarity 0.85)
- Self-Attention-Based Contextual Modulation Improves Neural System Identification (ICLR 2025 · 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.
