Research map / Image generation / Autoregressive and latent generation
PieNet: Personalized Image Enhancement Network
Han-Ul Kim, Young Jun Koh, Chang-Su Kim
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 |
|---|---|---|---|
| use personalized preference vectors instead of a single-style enhancer | PieNet · MIT-Adobe 5K multiple users | mPSNR: 24.28 vs 21.74 (DUPE) | Improved |
| test on all five photographers instead of C only | DUPE · MIT-Adobe 5K | PSNR: 23.61 -> 21.74 | Hurt |
| test on all five photographers instead of C only | PieNet · MIT-Adobe 5K | PSNR: 25.28 -> 24.28 | Hurt |
| train with supervised paired data rather than unpaired data | HDR/DPE/DUPE vs WB/D&R · MIT-Adobe 5K single user | PSNR: 18.36 -> 23.61 | Improved |
| add the perceptual loss based on the embedding network (ablation) | PieNet · MIT-Adobe 5K single user | PSNR | Improved |
| use triplet-loss preference vector scheme with many preferred images (Npref=2000) instead of feature averaging (ablation) | PieNet · Expanded MIT-Adobe 5K, 8 new users | mPSNR | Improved |
| use triplet-loss preference vector scheme with few preferred images instead of feature averaging (ablation) | PieNet · Expanded MIT-Adobe 5K, 8 new users | mPSNR | Hurt |
| use feature averaging with only 20 preferred images (ablation) | PieNet · Expanded MIT-Adobe 5K, 8 new users | mPSNR | No clear effect |
Closest papers by meaning
- Personalized Image Aesthetics (ICCV 2017 · similarity 0.84)
- Personalized Image Editing in Text-to-Image Diffusion Models via Collaborative Direct Preference Optimization (NeurIPS 2025 · similarity 0.84)
- StarEnhancer: Learning Real-Time and Style-Aware Image Enhancement (ICCV 2021 · similarity 0.83)
- Classification-Driven Dynamic Image Enhancement (CVPR 2018 · similarity 0.83)
- DreamSteerer: Enhancing Source Image Conditioned Editability using Personalized Diffusion Models (NeurIPS 2024 · similarity 0.81)
- Energy-Guided Optimization for Personalized Image Editing with Pretrained Text-to-Image Diffusion Models (AAAI 2025 · similarity 0.80)
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.
