Research map / Image generation / GANs
Toward Realistic Image Compositing With Adversarial Learning
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 |
|---|---|---|---|
| Use GCC-GAN instead of Alpha composition, Poisson blending, Deep harmonization, or Pix2Pix | COCO composite images with geometrically matched foregrounds · COCO human perception, single image | annotated as real: 4.1% (Alpha composition); 10.0% (Poisson blending); 8.6% (Deep harmonization); 10.2% (Pix2Pix) -> 11.0% | Helped |
| Use GCC-GAN instead of Alpha composition | COCO composite images with matched foreground and background · COCO human perception, paired comparison | GCC-GAN selected as more realistic: 82.5% | Helped |
| Use GCC-GAN instead of Poisson blending | COCO composite images with matched foreground and background · COCO human perception, paired comparison | GCC-GAN selected as more realistic: 67.3% | Helped |
| Use GCC-GAN instead of Deep harmonization | COCO composite images with matched foreground and background · COCO human perception, paired comparison | GCC-GAN selected as more realistic: 71.4% | Helped |
| Use GCC-GAN instead of Pix2Pix | COCO composite images with matched foreground and background · COCO human perception, paired comparison | GCC-GAN selected as more realistic: 56.7% | Helped |
| Use GCC-GAN instead of Alpha composition, Poisson blending, Deep harmonization, or Pix2Pix | COCO composite images evaluated by RGB-N · COCO manipulation detection | average RGB-N score: 75.4% (Alpha composition); 75.8% (Poisson blending); 77.0% (Deep harmonization); 69.1% (Pix2Pix) -> 63.7% | Helped |
| Add color correction alongside geometric transformation (ablation) | GCC-GAN on synthesized soda-can/table composites · Synthesized objects | plausibility of composite images | Helped |
| Add geometric transformation alongside color correction (ablation) | GCC-GAN on COCO composites with geometric mismatch · COCO geometric correction | plausibility of composite images | Helped |
Closest papers by meaning
- ST-GAN: Spatial Transformer Generative Adversarial Networks for Image Compositing (CVPR 2018 · similarity 0.88)
- Neural Photo Editing with Introspective Adversarial Networks (ICLR 2017 · similarity 0.87)
- Exploiting Spatial Dimensions of Latent in GAN for Real-Time Image Editing (CVPR 2021 · similarity 0.87)
- Drop the GAN: In Defense of Patches Nearest Neighbors As Single Image Generative Models (CVPR 2022 · similarity 0.87)
- Spatial Fusion GAN for Image Synthesis (CVPR 2019 · similarity 0.86)
- E$^2$GAN: Efficient Training of Efficient GANs for Image-to-Image Translation (ICML 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.