Research map / Label-efficient and robust learning / Semi-supervised learning
Contrastive Semi-Supervised Learning for Underwater Image Restoration via Reliable Bank
Read the paperCheck 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 |
|---|---|---|---|
| use the full Semi-UIR framework instead of supervised-only AIM-Net (ablation) | AIM-Net · testR | PSNR: 24.38 -> 24.59 | Helped |
| add mean-teacher consistency training without reliable bank or contrastive loss (ablation) | AIM-Net · testR | PSNR: 24.38 -> 23.11 | Hurt |
| add the reliable bank to mean-teacher training (ablation) | Semi-base · testR | PSNR: 23.11 -> 24.53 | Helped |
| add the reliable bank to mean-teacher training (ablation) | Semi-base · EUVP | MUSIQ: 42.48 -> 43.27 | Helped |
| add contrastive loss without a reliable bank (ablation) | Semi-base · EUVP | MUSIQ: 42.48 -> 46.59 | Helped |
| add contrastive loss without a reliable bank (ablation) | Semi-base · UIEB | MUSIQ: 40.33 -> 40.64 | No clear effect |
| use MUSIQ rather than NIMA to select reliable-bank outputs (ablation) | Semi-UIR · testR | PSNR: 23.88 -> 24.59 | Helped |
| use MUSIQ rather than PAQ2PIQ to select reliable-bank outputs (ablation) | Semi-UIR · testR | PSNR: 24.28 -> 24.59 | Helped |
| apply color jitter augmentation (ablation) | Semi-UIR · EUVP | MUSIQ: 46.06 -> 49.16 | Helped |
| apply Gaussian blur augmentation (ablation) | Semi-UIR · RUIE | MUSIQ: 31.14 -> 36.88 | Helped |
| apply grayscale augmentation (ablation) | Semi-UIR · Seathru | MUSIQ: 64.71 -> 65.19 | Helped |
| combine color jitter, Gaussian blur, and grayscale augmentations (ablation) | Semi-UIR · EUVP | MUSIQ: 46.06 -> 51.66 | Helped |
Closest papers by meaning
- Synergistic Multiscale Detail Refinement via Intrinsic Supervision for Underwater Image Enhancement (AAAI 2024 · similarity 0.86)
- Adaptive Dual-domain Learning for Underwater Image Enhancement (AAAI 2025 · similarity 0.86)
- Wavelet-based Fourier Information Interaction with Frequency Diffusion Adjustment for Underwater Image Restoration (CVPR 2024 · similarity 0.85)
- UVEB: A Large-scale Benchmark and Baseline Towards Real-World Underwater Video Enhancement (CVPR 2024 · similarity 0.84)
- Learning from History: Task-agnostic Model Contrastive Learning for Image Restoration (AAAI 2024 · similarity 0.84)
- Atlantis: Enabling Underwater Depth Estimation with Stable Diffusion (CVPR 2024 · similarity 0.83)
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.