Research map / Image generation / Autoregressive and latent generation
UCTGAN: Diverse Image Inpainting Based on Unsupervised Cross-Space Translation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 cross semantic attention (ablation) | UCTGAN · CelebA-HQ test, center mask | PSNR: 26.0223 -> 26.3833 | Helped |
| add cross semantic attention (ablation) | UCTGAN · CelebA-HQ test, center mask | SSIM: 0.8732 -> 0.8862 | Helped |
| add cross semantic attention (ablation) | UCTGAN · CelebA-HQ test, center mask | IS: 3.0011 -> 3.0127 | Helped |
| add cross semantic attention (ablation) | UCTGAN · CelebA-HQ test, center mask | MIS: 0.0174 -> 0.0178 | Helped |
| add cross semantic attention (ablation) | UCTGAN · CelebA-HQ test, center mask | l1 loss (%): 1.65 -> 1.51 | Helped |
| use instance-image guidance instead of standard normal noise (ablation) | UCTGAN · CelebA-HQ test, center mask | MIS: 0.0174 -> 0.0178 | Helped |
| use instance-image guidance instead of standard normal noise (ablation) | UCTGAN · CelebA-HQ test, center mask | PSNR: 25.9700 -> 26.3833 | Helped |
| add low-dimension loss Llrec (ablation) | UCTGAN · CelebA-HQ, 1K center-masked images | LPIPS(Iout): 0.017 -> 0.030 | Helped |
| add low-dimension loss Llrec (ablation) | UCTGAN · CelebA-HQ, 1K center-masked images | LPIPS(Iout(m)): 0.032 -> 0.092 | Helped |
| use instance-image guidance instead of standard normal noise (ablation) | UCTGAN · CelebA-HQ, 1K center-masked images | LPIPS(Iout(m)): 0.062 -> 0.092 | Helped |
Closest papers by meaning
- Homomorphic Latent Space Interpolation for Unpaired Image-To-Image Translation (CVPR 2019 · similarity 0.88)
- InOut: Diverse Image Outpainting via GAN Inversion (CVPR 2022 · similarity 0.88)
- Generative Image Inpainting With Contextual Attention (CVPR 2018 · similarity 0.87)
- Image Inpainting via Generative Multi-column Convolutional Neural Networks (NeurIPS 2018 · similarity 0.87)
- Learning Pyramid-Context Encoder Network for High-Quality Image Inpainting (CVPR 2019 · similarity 0.87)
- Unsupervised Image-to-Image Translation With Generative Prior (CVPR 2022 · 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.