Research map / Label-efficient and robust learning / Domain adaptation and generalization
On the Effectiveness of Image Rotation for Open Set Domain Adaptation
Silvia Bucci, Mohammad Reza Loghmani, Tatiana Tommasi
Read the paperCheck 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 |
|---|---|---|---|
| use rotation-recognition known/unknown separation instead of multi-binary classifier | ROS Stage I · Office-31 | AUC-ROC: 79.9 -> 91.5 | Improved |
| add center loss (ablation) | ROS Stage I · Office-31 | AUC-ROC: 89.9 -> 91.5 | Improved |
| add anchor image (relative rotation) (ablation) | ROS Stage I · Office-31 | AUC-ROC: 90.4 -> 91.5 | Improved |
| add rotation score to normality score (ablation) | ROS Stage I · Office-31 | AUC-ROC: 89.6 -> 91.5 | Improved |
| add entropy score to normality score (ablation) | ROS Stage I · Office-31 | AUC-ROC: 88.3 -> 91.5 | Improved |
| add center loss and anchor together (ablation) | ROS Stage I · Office-31 | AUC-ROC: 87.1 -> 91.5 | Improved |
| use rotation recognition instead of GRL domain discriminator in Stage II (ablation) | ROS Stage II · Office-31 | HOS: 85.1 -> 85.9 | Improved |
| use relative rotation with anchor in Stage II (ablation) | ROS Stage II · Office-31 | HOS: 84.9 -> 85.9 | Improved |
| add entropy loss in Stage II (without anchor) (ablation) | ROS Stage II · Office-31 | HOS: 85.0 vs 84.9 | No clear effect |
| apply rotation-based adaptation in Stage II (ablation) | ROS · Office-31 | cosine distance source-target: 0.188 -> 0.109 | Improved |
| use HOS rather than OS for evaluation | UAN · Office-Home | OS vs UNK: OS 72.5 vs UNK 0.0 | Mixed |
| increase openness (fewer known classes) | STA and OSBP · Office-Home | HOS | Hurt |
Closest papers by meaning
- Separate to Adapt: Open Set Domain Adaptation via Progressive Separation (CVPR 2019 · similarity 0.91)
- Adjustment and Alignment for Unbiased Open Set Domain Adaptation (CVPR 2023 · similarity 0.90)
- Open-Set Heterogeneous Domain Adaptation: Theoretical Analysis and Algorithm (AAAI 2025 · similarity 0.87)
- Open-Set Domain Adaptation for Semantic Segmentation (CVPR 2024 · similarity 0.87)
- COSDA: Counterfactual-based Susceptibility Risk Framework for Open-Set Domain Adaptation (ICML 2025 · similarity 0.86)
- Dynamic Target Distribution Estimation for Source-Free Open-Set Domain Adaptation (AAAI 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.
