Research map / Image generation / Image restoration
Towards Robust Blind Face Restoration with Codebook Lookup Transformer
Shangchen Zhou, Kelvin C.K. Chan, Chongyi Li, Chen Change Loy
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 the learned discrete codebook (vs plain encoder-decoder) (ablation) | CodeFormer · CelebA-Test | LPIPS: 0.420 -> 0.397 | Improved |
| add the codebook (vs plain encoder-decoder) (ablation) | CodeFormer · CelebA-Test | IDS: 0.47 -> 0.51 | Improved |
| predict codes with a linear/CNN head instead of nearest-neighbour matching (ablation) | CodeFormer · CelebA-Test | LPIPS: 0.397 -> 0.351 | Improved |
| use Transformer for code prediction instead of CNN (ablation) | CodeFormer · CelebA-Test | LPIPS: 0.351 -> 0.307 | Improved |
| use Transformer for code prediction instead of CNN (ablation) | CodeFormer · CelebA-Test | IDS: 0.55 -> 0.58 | Improved |
| fine-tune the decoder instead of keeping it fixed (ablation) | CodeFormer · CelebA-Test | LPIPS: 0.307 -> 0.379 | Hurt |
| set CFT coefficient w=1 instead of w=0 (ablation) | CodeFormer · CelebA-Test | LPIPS: 0.307 -> 0.297 | Improved |
| set CFT coefficient w=1 instead of w=0 (ablation) | CodeFormer · CelebA-Test | IDS: 0.58 -> 0.60 | Improved |
| increase codebook size N from 384 to 1024 (ablation) | codebook reconstruction (Stage I) · FFHQ | LPIPS: 0.202 -> 0.175 | Improved |
| increase codebook size N from 384 to 1024 (ablation) | codebook reconstruction (Stage I) · FFHQ | PSNR: 22.59 -> 23.41 | Improved |
| use multi-scale encoder features as Transformer input instead of single-scale (ablation) | CodeFormer · CelebA-Test | code prediction accuracy: 0.192 -> 0.188 | No clear effect |
| increase Transformer depth from 9 to 15 layers (ablation) | CodeFormer · CelebA-Test | code prediction accuracy: 0.192 -> 0.193 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GLEAN — GLEAN: Generative Latent Bank for Large-Factor Image Super-Resolution (CVPR 2021) · better on 5 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- RestoreFormer++: Towards Real-World Blind Face Restoration From Undegraded Key-Value Pairs (TPAMI 2023 · similarity 0.91)
- Blind Face Restoration under Extreme Conditions: Leveraging 3D-2D Prior Fusion for Superior Structural and Texture Recovery (AAAI 2024 · similarity 0.87)
- FaceMe: Robust Blind Face Restoration with Personal Identification (AAAI 2025 · similarity 0.86)
- Learning Degradation-unaware Representation with Prior-based Latent Transformations for Blind Face Restoration (CVPR 2024 · similarity 0.86)
- Switchable Token-Specific Codebook Quantization For Face Image Compression (NeurIPS 2025 · similarity 0.85)
- RetouchFormer: Semi-supervised High-Quality Face Retouching Transformer with Prior-Based Selective Self-Attention (AAAI 2024 · similarity 0.85)
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.
