Research map / Image generation / Autoregressive and latent generation
Pluralistic Image Completion
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 |
|---|---|---|---|
| Couple a reconstructive path to a sampled generative path using prior-conditional coupling (ablation) | PICNet · Center-masked images | LPIPS, mask-region output: 0.014 -> 0.088 | Helped |
| Replace dual-pipeline training with instance-blind training (ablation) | PICNet · Center-masked images | LPIPS, mask-region output: 0.088 -> 0.049 | Hurt |
| Use CVAE training instead of the dual-pipeline framework (ablation) | PICNet · Center-masked images | LPIPS, full output: 0.029 -> 0.004 | Hurt |
| Use instance-blind training instead of the dual-pipeline framework (ablation) | PICNet · Center-masked images | LPIPS, full output: 0.029 -> 0.015 | Hurt |
| Use BicycleGAN for center-mask completion instead of PICNet (ablation) | Pluralistic image completion · Center-masked images | LPIPS, mask-region output: 0.088 -> 0.060 | Hurt |
| Add short+long term attention to combine decoder self-attention with encoder context (ablation) | PICNet generator · Image completion, including large face masks | visual completion quality | Helped |
| Increase the KL loss weight (ablation) | PICNet · Generated missing regions | diversity | Helped |
| Increase the KL loss weight (ablation) | PICNet · Generated missing regions | appearance consistency with visible regions | Hurt |
| Use one discriminator step per generator step (ablation) | PICNet training · Image completion | overall results | Helped |
Closest papers by meaning
- Pluralistic Image Completion with Gaussian Mixture Models (NeurIPS 2022 · similarity 0.90)
- Large Scale Image Completion via Co-Modulated Generative Adversarial Networks (ICLR 2021 · similarity 0.89)
- MaGIC: Multi-modality Guided Image Completion (ICLR 2024 · similarity 0.86)
- Don't Look into the Dark: Latent Codes for Pluralistic Image Inpainting (CVPR 2024 · similarity 0.85)
- Large Occluded Human Image Completion via Image-Prior Cooperating (AAAI 2024 · similarity 0.85)
- Completion as Enhancement: A Degradation-Aware Selective Image Guided Network for Depth Completion (CVPR 2025 · similarity 0.83)
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.