Research map / Label-efficient and robust learning / Domain adaptation and generalization
Geometric and Textural Augmentation for Domain Gap Reduction
Xiao-Chang Liu, Yong-Liang Yang, Peter Hall
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 geometric and texture augmentation (ablation) | ResNet-18 · PACS multi-source | accuracy: 79.51 -> 86.56 | Improved |
| Add texture augmentation (ablation) | ResNet-18 with geometric augmentation · PACS multi-source | accuracy: 83.05 -> 86.56 | Improved |
| Add geometric augmentation (ablation) | ResNet-18 with texture augmentation · PACS multi-source | accuracy: 86.07 -> 86.56 | Improved |
| Add geometric and texture augmentation (ablation) | ResNet-50 · PACS multi-source | accuracy: 82.02 -> 89.17 | Improved |
| Add geometric and texture augmentation (ablation) | the proposed method · Office-Home multi-source | accuracy: 64.01 -> 66.75 | Improved |
| Add geometric augmentation (ablation) | the proposed method with texture augmentation · Office-Home multi-source | accuracy: 65.12 -> 66.75 | Improved |
| Add texture augmentation (ablation) | the proposed method with geometric augmentation · Office-Home multi-source | accuracy: 66.60 -> 66.75 | Improved |
| Add geometric and texture augmentation (ablation) | the proposed method · Digits-DG multi-source | accuracy: 75.0 -> 82.6 | Improved |
| Add geometric augmentation (ablation) | the proposed method with texture augmentation · Digits-DG multi-source | accuracy: 78.6 -> 82.6 | Improved |
| Add texture augmentation (ablation) | the proposed method with geometric augmentation · Digits-DG multi-source | accuracy: 80.3 -> 82.6 | Improved |
| Use geometric and texture augmentation | the proposed method · PACS single-source | accuracy: +3% | Improved |
| Use geometric and texture augmentation | the proposed method · PACS multi-source | accuracy: 86.56 vs 84.51 | Improved |
Closest papers by meaning
- Shift from Texture-bias to Shape-bias: Edge Deformation-based Augmentation for Robust Object Recognition (ICCV 2023 · similarity 0.88)
- Learning To Warp for Style Transfer (CVPR 2021 · similarity 0.88)
- Image Style Transfer Using Convolutional Neural Networks (CVPR 2016 · similarity 0.88)
- Reducing Domain Gap by Reducing Style Bias (CVPR 2021 · similarity 0.88)
- 3DStyleNet: Creating 3D Shapes With Geometric and Texture Style Variations (ICCV 2021 · similarity 0.87)
- ImageNet-trained CNNs are not biased towards texture: Revisiting feature reliance through controlled suppression (NeurIPS 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.
