Research map / Image generation / Image restoration
Single Image Dehazing via Conditional Generative Adversarial Network
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 cGAN with perceptual, pixel-wise, and gradient-prior losses instead of original cGAN | Single-image dehazing cGAN · Synthetic testing dataset | PSNR: 31.09 -> 33.61 | Helped |
| Use the full cGAN with perceptual, pixel-wise, and gradient-prior losses instead of original cGAN | Single-image dehazing cGAN · Synthetic testing dataset | SSIM: 0.802 -> 0.915 | Helped |
| Add the perceptual loss to adversarial, pixel-wise, and gradient-prior losses (ablation) | Single-image dehazing cGAN · Synthetic testing dataset | SSIM: 0.883 -> 0.915 | Helped |
| Add the perceptual loss to adversarial, pixel-wise, and gradient-prior losses (ablation) | Single-image dehazing cGAN · Synthetic testing dataset | PSNR: 33.81 -> 33.61 | Hurt |
| Add the L1-regularized gradient prior to adversarial and pixel-wise losses (ablation) | Single-image dehazing cGAN · Synthetic testing dataset | PSNR: 31.76 -> 33.81 | Helped |
| Add the L1-regularized gradient prior to adversarial and pixel-wise losses (ablation) | Single-image dehazing cGAN · Synthetic testing dataset | SSIM: 0.887 -> 0.883 | No clear effect |
| Add the perceptual loss to adversarial and pixel-wise L1 losses, without the gradient prior (ablation) | Single-image dehazing cGAN · Synthetic testing dataset | SSIM: 0.887 -> 0.908 | Helped |
| Add the perceptual loss to adversarial and pixel-wise L1 losses, without the gradient prior (ablation) | Single-image dehazing cGAN · Synthetic testing dataset | PSNR: 31.76 -> 32.82 | Helped |
| Add the L1-regularized gradient prior to adversarial and perceptual losses (ablation) | Single-image dehazing cGAN · Synthetic testing dataset | PSNR: 31.19 -> 33.61 | Helped |
| Add the L1-regularized gradient prior to adversarial and perceptual losses (ablation) | Single-image dehazing cGAN · Synthetic testing dataset | SSIM: 0.8966 -> 0.915 | Helped |
| Use summation rather than concatenation for symmetric generator skip connections (ablation) | Single-image dehazing generator · Synthetic testing dataset | PSNR: 33.45dB -> 33.61dB | Helped |
| Use summation rather than concatenation for symmetric generator skip connections (ablation) | Single-image dehazing generator · Synthetic testing dataset | SSIM: 0.9092 -> 0.9152 | Helped |
Closest papers by meaning
- FD-GAN: Generative Adversarial Networks with Fusion-Discriminator for Single Image Dehazing (AAAI 2020 · similarity 0.92)
- Dehaze-RetinexGAN: Real-World Image Dehazing via Retinex-based Generative Adversarial Network (AAAI 2025 · similarity 0.91)
- Deblurring by Realistic Blurring (CVPR 2020 · similarity 0.88)
- Fully Zero-Shot Image Dehazing (ICML 2026 · similarity 0.88)
- Towards Perceptual Image Dehazing by Physics-Based Disentanglement and Adversarial Training (AAAI 2018 · similarity 0.88)
- Learning Hazing to Dehazing: Towards Realistic Haze Generation for Real-World Image Dehazing (CVPR 2025 · similarity 0.85)
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.