Research map / Image generation / GANs
Unsupervised Class-Specific Deblurring
Thekke Madam Nimisha, Kumar Sunil, A. N. Rajagopalan
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 reblurring CNN module to GAN-only training (ablation) | unsupervised GAN deblurrer · Face deblurring (qualitative, Fig. 3) | visual quality | Improved |
| add scale-space gradient module on top of GAN plus reblur (ablation) | unsupervised GAN deblurrer · Face deblurring (qualitative, Fig. 3) | visual quality | Improved |
| use reblur-and-gradient self-guided GAN instead of CycleGAN | unsupervised deblurring · Face deblurring (synthetic, CelebA) | PSNR (dB): 8.93 -> 22.80 | Improved |
| use reblur-and-gradient self-guided GAN instead of CycleGAN | unsupervised deblurring · Text deblurring (synthetic) | PSNR (dB): 13.19 -> 23.22 | Improved |
| use reblur-and-gradient self-guided GAN instead of CycleGAN | unsupervised deblurring · Checkerboard deblurring (synthetic) | PSNR (dB): 21.92 -> 20.61 | Hurt |
| unsupervised training instead of supervised Nah et al. | deblurring network · Face deblurring (synthetic, CelebA) | PSNR (dB): 24.12 -> 22.80 | Hurt |
| unsupervised training instead of supervised Nah et al. | deblurring network · Text deblurring (synthetic) | PSNR (dB): 18.72 -> 23.22 | Improved |
| unsupervised training instead of supervised text-specific Hradis et al. | deblurring network · Text deblurring (synthetic) | PSNR (dB): 24.28 -> 23.22 | Hurt |
| unsupervised training instead of supervised Hradis et al. | deblurring network · Text deblurring (synthetic) | CER: 0.0891 -> 0.126 | Hurt |
| use unsupervised network instead of conventional Pan et al. [30] text prior | deblurring · Text human survey | user preference: 97% | Improved |
| use unsupervised network instead of supervised Hradis et al. [12] | deblurring · Text human survey | user preference: 43% vs 57% | Hurt |
| use unsupervised network instead of exemplar-based Pan et al. [29] (with GT shown) | face deblurring · Face human survey | user preference: 81% | Improved |
Closest papers by meaning
- Deblurring by Realistic Blurring (CVPR 2020 · similarity 0.90)
- Unsupervised Domain-Specific Deblurring via Disentangled Representations (CVPR 2019 · similarity 0.90)
- Deep Image Deblurring: A Survey (IJCV 2022 · similarity 0.90)
- Unsupervised Blind Image Deblurring Based on Self-Enhancement (CVPR 2024 · similarity 0.89)
- Dehaze-RetinexGAN: Real-World Image Dehazing via Retinex-based Generative Adversarial Network (AAAI 2025 · similarity 0.88)
- Cross-Scale Self-Supervised Blind Image Deblurring via Implicit Neural Representation (NeurIPS 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.
