Research map / Image generation / Image restoration
CT2: Colorization Transformer via Color Tokens
Shuchen Weng, Jimeng Sun, Yu Li, Si Li, Boxin Shi
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 luminance-selecting module (ablation) | CT2 · ImageNet val (first 5k) | FID: 7.51 -> 5.51 | Improved |
| add luminance-selecting module (ablation) | CT2 · ImageNet val (first 5k) | PSNR: 20.99 -> 23.50 | Improved |
| add luminance-selecting module (ablation) | CT2 · ImageNet val (first 5k) | SSIM: 0.82 -> 0.92 | Improved |
| add color attention (ablation) | CT2 · ImageNet val (first 5k) | FID: 7.76 -> 5.51 | Improved |
| add color attention (ablation) | CT2 · ImageNet val (first 5k) | PSNR: 20.93 -> 23.50 | Improved |
| add color attention (ablation) | CT2 · ImageNet val (first 5k) | LPIPS: 0.27 -> 0.19 | Improved |
| add color query module (ablation) | CT2 · ImageNet val (first 5k) | FID: 8.87 -> 5.51 | Improved |
| add color query module (ablation) | CT2 · ImageNet val (first 5k) | PSNR: 21.70 -> 23.50 | Improved |
| add color query module (ablation) | CT2 · ImageNet val (first 5k) | LPIPS: 0.23 -> 0.19 | Improved |
| add luminance-selecting module (ablation) | CT2 · ImageNet val (first 5k) | colorfulness: 41.56 -> 38.48 | Hurt |
| add color attention (ablation) | CT2 · ImageNet val (first 5k) | colorfulness: 39.53 -> 38.48 | Hurt |
| add color query module (ablation) | CT2 · ImageNet val (first 5k) | colorfulness: 39.51 -> 38.48 | Hurt |
Closest papers by meaning
- Colorization Transformer (ICLR 2021 · similarity 0.92)
- ColorFormer: Image Colorization via Color Memory Assisted Hybrid-Attention Transformer (ECCV 2022 · similarity 0.90)
- Colorful Image Colorization (ECCV 2016 · similarity 0.89)
- Recolour What Matters: Region-Aware Colour Editing via Token-Level Diffusion (ECCV 2026 · similarity 0.89)
- Pixelated Semantic Colorization (IJCV 2019 · similarity 0.88)
- Automatic Controllable Colorization via Imagination (CVPR 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.
