Research map / Architectures and efficiency / Neural architecture search
ISNAS-DIP: Image-Specific Neural Architecture Search for Deep Image Prior
Metin Ersin Arican, Ozgur Kara, Gustav Bredell, Ender Konukoglu
Check 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 |
|---|---|---|---|
| Select image-specific architectures using PSD DB Strip MSE | DIP · BM3D denoising (σ=25) | PSNR: 27.87 -> 28.39 | Improved |
| Select image-specific architectures using PSD DB Strip MSE | DIP · Set12 denoising (σ=25) | PSNR: 27.92 -> 28.06 | Improved |
| Select image-specific architectures using PSD DB Strip MSE | DIP · CBM3D denoising (σ=25) | PSNR: 28.93 -> 30.36 | Improved |
| Select image-specific architectures using PSD DB MSE | DIP · BM3D inpainting | PSNR: 31.04 -> 32.90 | Improved |
| Select image-specific architectures using PSD DB MSE | DIP · Set12 inpainting | PSNR: 31.00 -> 32.22 | Improved |
| Use image-specific PSD metrics for architecture selection | DIP · Super-resolution, Set5 ×2 and ×4; Set14 ×2, ×4, and ×8 | PSNR | Hurt |
| Restrict the PSD comparison to a frequency strip, avoiding the full PSD (ablation) | ISNAS-DIP · Denoising | PSNR: 24.31 -> 28.39; 23.77 -> 28.06; 30.01 -> 30.36 | Improved |
| Use PSD DB Strip MSE rather than whole-PSD DB MSE (ablation) | ISNAS-DIP · Super-resolution | PSNR: 34.83 -> 34.72; 30.22 -> 30.05; 23.97 -> 23.90 | Mixed |
| Use PSD DB Strip MSE rather than PSD DB MSE (ablation) | ISNAS-DIP · BM3D inpainting | PSNR: 32.90 -> 32.32 | Hurt |
| Use resized averaging selection rather than full-sized averaging selection (ablation) | PSD DB Strip MSE selection · Denoising | PSNR: 27.66 -> 28.39; 27.40 -> 28.06; 29.28 -> 30.36 | Hurt |
| Use resized averaging selection rather than full-sized averaging selection (ablation) | PSD DB Strip MSE selection · Inpainting | PSNR: 33.01 -> 32.32; 32.17 -> 31.68 | Mixed |
| Select candidates by PSD DB Strip MSE rather than random search-space selection (ablation) | ISNAS-DIP · Denoising | PSNR: 23.20 -> 27.66; 22.66 -> 27.40; 27.86 -> 29.28 | Improved |
Closest papers by meaning
- NAS-DIP: Learning Deep Image Prior with Neural Architecture Search (ECCV 2020 · similarity 0.94)
- Memory-Efficient Hierarchical Neural Architecture Search for Image Denoising (CVPR 2020 · similarity 0.90)
- Memory-Efficient Hierarchical Neural Architecture Search for Image Restoration (IJCV 2021 · similarity 0.90)
- CLEARER: Multi-Scale Neural Architecture Search for Image Restoration (NeurIPS 2020 · similarity 0.90)
- L-SWAG: Layer-Sample Wise Activation with Gradients Information for Zero-Shot NAS on Vision Transformers (CVPR 2025 · similarity 0.88)
- Training-free Neural Architecture Search through Variance of Knowledge of Deep Network Weights (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.
