Research map / Video / Video generation
Deep Learning-Based Image and Video Inpainting: A Survey
Weize Quan, J. Chen, Yanli Liu, Dong-ming Yan A, Peter Wonka
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 style loss (ablation) | inpainting network with pixel loss · CelebA-HQ, mask 20%-30% | FID: 3.31 -> 2.25 | Improved |
| add style loss (ablation) | inpainting network with pixel loss · CelebA-HQ, mask 20%-30% | PSNR: 30.76 -> 30.24 | Hurt |
| add perceptual loss (VGG16) (ablation) | inpainting network with pixel loss · CelebA-HQ, mask 20%-30% | FID: 3.31 -> 3.20 | Improved |
| add perceptual loss (VGG16) (ablation) | inpainting network with pixel loss · CelebA-HQ, mask 20%-30% | LPIPS: 0.063 -> 0.057 | Improved |
| add perceptual loss from segmentation ResNet50 (ablation) | inpainting network with pixel loss · CelebA-HQ, mask 20%-30% | LPIPS: 0.063 -> 0.052 | Improved |
| add LSGAN adversarial loss (ablation) | inpainting network with pixel loss · CelebA-HQ, mask 20%-30% | LPIPS: 0.063 -> 0.054 | Improved |
| add LSGAN adversarial loss (ablation) | inpainting network with pixel loss · CelebA-HQ, mask 20%-30% | FID: 3.31 -> 3.34 | No clear effect |
| combine perceptual and style loss instead of style alone (ablation) | inpainting network with pixel loss · CelebA-HQ, mask 20%-30% | FID: 2.25 -> 2.26 | No clear effect |
| add style loss (ablation) | inpainting network with pixel loss · Paris StreetView, mask 20%-30% | FID: 26.55 -> 19.65 | Improved |
| use LGNet instead of MAT | image inpainting · Places2, mask 20%-30% | PSNR: 26.41 -> 28.18 | Improved |
| use diffusion-based RePaint instead of LGNet | image inpainting · CelebA-HQ, mask 20%-30% | PSNR: 30.54 -> 28.38 | Hurt |
Closest papers by meaning
- Deep Video Inpainting (CVPR 2019 · similarity 0.92)
- An Internal Learning Approach to Video Inpainting (ICCV 2019 · similarity 0.91)
- Contextual-based Image Inpainting: Infer, Match, and Translate (ECCV 2018 · similarity 0.90)
- Generative Image Inpainting With Contextual Attention (CVPR 2018 · similarity 0.90)
- Copy-and-Paste Networks for Deep Video Inpainting (ICCV 2019 · similarity 0.90)
- High-Resolution Image Inpainting with Iterative Confidence Feedback and Guided Upsampling (ECCV 2020 · similarity 0.89)
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.
