Research map / Image generation / Autoregressive and latent generation
Unleashing the Potential of the Semantic Latent Space in Diffusion Models for Image Dehazing
Zizheng Yang, Hu Yu, Bing Li, Jinghao Zhang, Jie Huang, Feng Zhao
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 CIM (h-space content integration at t1) to a baseline (ablation) | DiffLI2D · SOTS indoor | PSNR: 36.65 -> 38.44 | Improved |
| add HAE (h-space haze guidance at t2) to a baseline (ablation) | DiffLI2D · SOTS indoor | PSNR: 36.65 -> 38.21 | Improved |
| use both CIM and HAE (ablation) | DiffLI2D · SOTS indoor | PSNR: 38.44 -> 40.33 | Improved |
| use diffusion h-space features instead of encoder features (Layer-E) (ablation) | DiffLI2D · SOTS indoor | PSNR: 36.52 -> 40.33 | Improved |
| use diffusion h-space features instead of decoder features (Layer-D) (ablation) | DiffLI2D · SOTS indoor | PSNR: 39.10 -> 40.33 | Improved |
| use h-space features of a diffusion model instead of VGG16 features (ablation) | DiffLI2D · SOTS indoor | PSNR: 27.83 -> 40.33 | Improved |
| use h-space features instead of ResNet50 features (ablation) | DiffLI2D · SOTS indoor | PSNR: 25.49 -> 40.33 | Improved |
| set t2 around 500 instead of 100 or 600 (ablation) | DiffLI2D · SOTS indoor | PSNR: +1.99dB vs t2=100; +2.07dB vs t2=600 | Improved |
| use very large time-steps (t1 or t2 = 1000) (ablation) | DiffLI2D · SOTS indoor | PSNR | Hurt |
| set t2 to 800 instead of 600 for low-light enhancement (ablation) | DiffLI2D · LOLv1 | PSNR: 23.30 -> 21.48 | Hurt |
| apply DiffLI2D to low-light enhancement | DiffLI2D · LOLv1 | PSNR: 20.61 -> 23.30 vs Restormer | Improved |
| DiffLI2D vs Dehamer on NH-HAZE | DiffLI2D · NH-HAZE | PSNR: 20.29 vs 20.66 | Mixed |
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.
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · better on 3 of 3 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
- Learning Hazing to Dehazing: Towards Realistic Haze Generation for Real-World Image Dehazing (CVPR 2025 · similarity 0.91)
- Exploiting Diffusion Prior for Real-World Image Dehazing with Unpaired Training (AAAI 2025 · similarity 0.91)
- Unsupervised Region-Based Image Editing of Denoising Diffusion Models (AAAI 2025 · similarity 0.90)
- Simpler Diffusion: 1.5 FID on ImageNet512 with Pixel-space Diffusion (CVPR 2025 · similarity 0.90)
- DeblurDiff: Real-Word Image Deblurring with Generative Diffusion Models (NeurIPS 2025 · similarity 0.89)
- CleanDIFT: Diffusion Features without Noise (CVPR 2025 · similarity 0.88)
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.
