Research map / Image generation / GANs
HardGAN: A Haze-Aware Representation Distillation GAN for Single Image Dehazing
Qili Deng, Ziling Huang, Chung-Chi Tsai, Chia-Wen Lin
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 |
|---|---|---|---|
| add local spatial structure preservation (L) (ablation) | HardGAN baseline · SOTS Outdoor | PSNR: 30.78 -> 33.54 | Improved |
| add local spatial structure preservation (L) (ablation) | HardGAN baseline · SOTS Indoor | PSNR: 32.25 -> 35.20 | Improved |
| add global atmospheric brightness control (G) (ablation) | HardGAN baseline · SOTS Outdoor | PSNR: 30.78 -> 33.77 | Improved |
| add global atmospheric brightness control (G) (ablation) | HardGAN baseline · SOTS Indoor | PSNR: 32.25 -> 35.48 | Improved |
| add global brightness control (G) on top of local preservation (ablation) | HardGAN with L · SOTS Outdoor | PSNR: 33.54 -> 34.34 | Improved |
| add local preservation (L) on top of global control (ablation) | HardGAN with G · SOTS Indoor | PSNR: 35.48 -> 36.56 | Improved |
| add both L and G (ablation) | HardGAN baseline · SOTS Outdoor | LPIPS: 0.025 -> 0.010 | Improved |
| add both L and G (ablation) | HardGAN baseline · SOTS Indoor | SSIM: 0.9838 -> 0.9905 | Improved |
| add local preservation (L) (ablation) | HardGAN baseline · SOTS Indoor | LPIPS: 0.220 -> 0.222 | No clear effect |
| use HardGAN instead of FFANet | FFANet · SOTS Indoor | PSNR: 36.39 -> 36.56 | Improved |
| use HardGAN instead of FFANet | FFANet · SOTS Outdoor | PSNR: 33.57 -> 34.34 | Improved |
| use HardGAN instead of best challenge entry | ECNU-Trident · NH-HAZE | PSNR: 21.41 -> 21.70 | Improved |
Later papers that compared with this method
1 later accepted paper reports HardGAN as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Self-Augmented Unpaired Image Dehazing via Density and Depth Decomposition (CVPR 2022 · better on 3 of 5 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FFANet — FFA-Net: Feature Fusion Attention Network for Single Image Dehazing (AAAI 2020) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- FD-GAN: Generative Adversarial Networks with Fusion-Discriminator for Single Image Dehazing (AAAI 2020 · similarity 0.91)
- Single Image Dehazing via Conditional Generative Adversarial Network (CVPR 2018 · similarity 0.91)
- Dehaze-RetinexGAN: Real-World Image Dehazing via Retinex-based Generative Adversarial Network (AAAI 2025 · similarity 0.90)
- Towards Domain Invariant Single Image Dehazing (AAAI 2021 · similarity 0.88)
- Learning Hazing to Dehazing: Towards Realistic Haze Generation for Real-World Image Dehazing (CVPR 2025 · similarity 0.86)
- PromptHaze: Prompting Real-world Dehazing via Depth Anything Model (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.
