Research map / Image generation / GANs
Mind the Gap: Domain Gap Control for Single Shot Domain Adaptation for Generative Adversarial Networks
Read the paperCheck 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 II2S to find a domain-A analog of the target reference instead of using the domain-A mean (ablation) | Mind the Gap · One-shot domain adaptation | visual quality and source-identity preservation | Helped |
| Add L_ref_clip to align the generated reference image with the target's CLIP embedding (ablation) | Mind the Gap · One-shot domain adaptation | reference semantic alignment | Helped |
| Add L_clip_within to preserve semantic changes within the source domain (ablation) | Mind the Gap · One-shot domain adaptation | preservation of within-domain variation | Helped |
| Add style mixing to move output styles toward the reference (ablation) | Mind the Gap · One-shot domain adaptation | visual similarity to reference style | Helped |
| Use Mind the Gap instead of StyleGAN-NADA | StyleGAN2 domain adaptation · Amazon Mechanical Turk user study | preference for image quality: 73% | Helped |
| Use Mind the Gap instead of Few-shot | StyleGAN2 domain adaptation · Amazon Mechanical Turk user study | preference for image quality: 77% | Helped |
| Use Mind the Gap instead of StyleGAN-NADA | StyleGAN2 domain adaptation · Amazon Mechanical Turk user study | preference for source-image similarity: 80% | Helped |
| Use Mind the Gap instead of Few-shot | StyleGAN2 domain adaptation · Amazon Mechanical Turk user study | preference for source-image similarity: 91% | Helped |
| Increase style-mixing parameter alpha | Mind the Gap · Style control | reference-style influence | Helped |
Closest papers by meaning
- Generalized One-shot Domain Adaptation of Generative Adversarial Networks (NeurIPS 2022 · similarity 0.91)
- Towards Diverse and Faithful One-shot Adaption of Generative Adversarial Networks (NeurIPS 2022 · similarity 0.91)
- HyperDomainNet: Universal Domain Adaptation for Generative Adversarial Networks (NeurIPS 2022 · similarity 0.89)
- Unsupervised Pixel-Level Domain Adaptation With Generative Adversarial Networks (CVPR 2017 · similarity 0.88)
- Generate to Adapt: Aligning Domains Using Generative Adversarial Networks (CVPR 2018 · similarity 0.87)
- Few Shot Generative Model Adaption via Relaxed Spatial Structural Alignment (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.