Research map / Image generation / GANs
WaveFill: A Wavelet-Based Generation Network for Image Inpainting
Yingchen Yu, Fangneng Zhan, Shijian Lu, Jianxiong Pan, Feiying Ma, Xuansong Xie, Chunyan Miao
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 wavelet decomposition instead of spatial domain (ablation) | Spatial + Concat baseline · Paris StreetView val (100, irregular masks) | FID: 33.95 -> 32.73 | Improved |
| use wavelet decomposition instead of spatial domain (ablation) | Spatial + Concat baseline · Paris StreetView val (100, irregular masks) | PSNR: 28.37 -> 28.46 | Improved |
| use DCT instead of wavelet for frequency split (ablation) | Wavelet + Concat · Paris StreetView val (100, irregular masks) | FID: 32.73 -> 100.93 | Hurt |
| use DCT instead of wavelet for frequency split (ablation) | Wavelet + Concat · Paris StreetView val (100, irregular masks) | PSNR: 28.46 -> 23.56 | Hurt |
| replace concatenation with SPADE fusion (ablation) | Wavelet + Concat · Paris StreetView val (100, irregular masks) | FID: 32.73 -> 32.14 | Improved |
| replace SPADE with FRAN fusion (ablation) | Wavelet + SPADE · Paris StreetView val (100, irregular masks) | FID: 32.14 -> 31.02 | Improved |
| replace concatenation with FRAN fusion (ablation) | Wavelet + Concat · Paris StreetView val (100, irregular masks) | FID: 32.73 -> 31.02 | Improved |
| replace concatenation with FRAN fusion (ablation) | Wavelet + Concat · Paris StreetView val (100, irregular masks) | PSNR: 28.46 -> 28.94 | Improved |
| replace SPADE with FRAN fusion (ablation) | Wavelet + SPADE · Paris StreetView val (100, irregular masks) | PSNR: 28.81 -> 28.94 | Improved |
| replace concatenation with FRAN fusion (ablation) | Wavelet + Concat · Paris StreetView val (100, irregular masks) | SSIM: 0.899 -> 0.904 | Improved |
| full WaveFill vs GC (DeepFill v2) on square masks | WaveFill · CelebA-HQ val (2000, 128x128 central square mask) | FID: 7.39 -> 6.48 | Improved |
| full WaveFill vs MEDFE at 40-50% irregular masks | WaveFill · Places2 val (irregular masks 40-50%) | FID: 28.98 -> 12.68 | Improved |
Later papers that compared with this method
1 later accepted paper reports WaveFill as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Generative Image Inpainting with Segmentation Confusion Adversarial Training and Contrastive Learning (AAAI 2023 · better on 4 of 4 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
- WaveFormer: Wavelet Transformer for Noise-Robust Video Inpainting (AAAI 2024 · similarity 0.87)
- High-Fidelity Image Inpainting with Multimodal Guided GAN Inversion (IJCV 2025 · similarity 0.86)
- High-Fidelity Image Inpainting with GAN Inversion (ECCV 2022 · similarity 0.86)
- InOut: Diverse Image Outpainting via GAN Inversion (CVPR 2022 · similarity 0.86)
- TurboFill: Adapting Few-step Text-to-image Model for Fast Image Inpainting (CVPR 2025 · similarity 0.85)
- Image Inpainting via Iteratively Decoupled Probabilistic Modeling (ICLR 2024 · similarity 0.84)
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.
