Research map / Image generation / Super-resolution
EnhanceNet: Single Image Super-Resolution Through Automated Texture Synthesis
Mehdi S. M. Sajjadi, Bernhard Scholkopf, Michael Hirsch
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 adversarial loss to MSE training (ablation) | ENet-E · Set5 (4x) | PSNR: 31.74 -> 28.15 | Hurt |
| add adversarial loss to perceptual loss (ablation) | ENet-P · Set14 (4x) | PSNR: 25.64 -> 24.93 | Hurt |
| add texture matching loss to perceptual plus adversarial loss (ablation) | ENet-PA · Set5 (4x) | PSNR: 27.20 -> 28.56 | Improved |
| add texture matching loss to perceptual plus adversarial loss (ablation) | ENet-PA · Urban100 (4x) | PSNR: 22.51 -> 23.54 | Improved |
| add texture loss to MSE plus adversarial loss (ablation) | ENet-EA · BSD100 (4x) | PSNR: 25.71 -> 25.97 | Improved |
| replace MSE with perceptual loss (ablation) | ENet-E · Set5 (4x) | PSNR: 31.74 -> 28.28 | Hurt |
| add adversarial loss to perceptual loss (ablation) | ENet-P · ImageNet ResNet-50 recognition | top-1 error: 0.449 -> 0.399 | Improved |
| add adversarial loss to MSE training (ablation) | ENet-E · ImageNet ResNet-50 recognition | top-1 error: 0.449 -> 0.407 | Improved |
| train with perceptual+adversarial+texture loss instead of MSE (ablation) | ENet-E · ImageNet ResNet-50 recognition | confidence: 0.754 -> 0.797 | Improved |
| use ENet-PAT instead of ENet-E in user study (ablation) | ENet-E · ImageNet user study | preference: 91.0% | Improved |
| train with MSE loss instead of bicubic baseline | Bicubic · Set5 (4x) | PSNR: 28.42 -> 31.74 | Improved |
| use ENet-E instead of VDSR | VDSR · Set5 (4x) | PSNR: 31.35 -> 31.74 | Improved |
Closest papers by meaning
- Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network (CVPR 2017 · similarity 0.92)
- SRFeat: Single Image Super-Resolution with Feature Discrimination (ECCV 2018 · similarity 0.91)
- SeD: Semantic-Aware Discriminator for Image Super-Resolution (CVPR 2024 · similarity 0.89)
- Exploring Semantic Feature Discrimination for Perceptual Image Super-Resolution and Opinion-Unaware No-Reference Image Quality Assessment (CVPR 2025 · similarity 0.89)
- Noise-Started One-Step Image Super-Resolution via LR-Conditioned SplitMeanFlow and GAN Refinement (NeurIPS 2026 · similarity 0.88)
- Auto-Encoded Supervision for Perceptual Image Super-Resolution (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.
