Research map / Label-efficient and robust learning / Semi-supervised learning
Incorporating Semi-Supervised and Positive-Unlabeled Learning for Boosting Full Reference Image Quality Assessment
Yue Cao, Zhaolin Wan, Dongwei Ren, Zifei Yan, Wangmeng Zuo
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 PU learning to SSL to filter outlier unlabeled images (ablation) | Proposed FR-IQA network · PIPAL | SRCC: 0.870 -> 0.874 | Improved |
| Add PU learning to SSL to filter outlier unlabeled images (ablation) | Proposed FR-IQA network · KADID-10k | SRCC: 0.958 -> 0.961 | Improved |
| Use unlabeled ESRGAN synthesis without PU outlier filtering (ablation) | Proposed FR-IQA network · KADID-10k | SRCC: 0.944 -> 0.932 | Hurt |
| Add DAB (ablation) | Proposed FR-IQA network · PIPAL | SRCC: 0.824 -> 0.837 | Improved |
| Add spatial attention (ablation) | Proposed FR-IQA network · PIPAL | SRCC: 0.857 -> 0.868 | Improved |
| Add LocalSW (ablation) | Proposed FR-IQA network · PIPAL | SRCC: 0.849 -> 0.868 | Improved |
| Replace global sliced Wasserstein with local distance patches (ablation) | Proposed FR-IQA network · PIPAL | SRCC: 0.755 -> 0.868 | Improved |
| Compute spatial attention from reference features rather than distortion features (ablation) | Proposed FR-IQA network · PIPAL Val. | PLCC: 0.861 -> 0.868 | Improved |
| Use spatial attention only instead of channel attention only (ablation) | Proposed FR-IQA network · PIPAL | SRCC: 0.840 -> 0.868 | Improved |
| Set τmin to 0.5 rather than 0.4 or 0.6 (ablation) | JSPL · PIPAL | SRCC: 0.870; 0.874; 0.872 | Improved |
| Use JSPL rather than supervised learning alone (ablation) | Proposed FR-IQA network · LIVE | SRCC: 0.970 -> 0.980 | Improved |
| Use JSPL rather than SSL with ESRGAN unlabeled data (ablation) | IQT · PIPAL Val., trained on KADID-10k and ESRGAN Synthesis | SRCC: 0.662 -> 0.783 | Improved |
Closest papers by meaning
- Semi-Supervised Blind Image Quality Assessment through Knowledge Distillation and Incremental Learning (AAAI 2024 · similarity 0.89)
- Semi-Supervised Blind Quality Assessment with Confidence-quantifiable Pseudo-label Learning for Authentic Images (ICML 2025 · similarity 0.88)
- Adaptive Image Quality Assessment via Teaching Large Multimodal Model to Compare (NeurIPS 2024 · similarity 0.88)
- Toward Generalized Image Quality Assessment: Relaxing the Perfect Reference Quality Assumption (CVPR 2025 · similarity 0.88)
- Adaptive Feature Selection for No-Reference Image Quality Assessment by Mitigating Semantic Noise Sensitivity (ICML 2024 · similarity 0.87)
- Transformer-Based No-Reference Image Quality Assessment via Supervised Contrastive Learning (AAAI 2024 · similarity 0.87)
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.
