Research map / Image generation / GANs
Sketch Your Own GAN
Sheng-Yu Wang, David Bau, Jun-Yan Zhu
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 cross-domain sketch adversarial loss with image regularization (Full w/o aug.) versus Chamfer baseline | StyleGAN2 horse rider · Horse rider | FID: 47.04 -> 27.50 | Improved |
| Full (w/o aug.) versus Chamfer baseline | StyleGAN2 standing cat · Standing cat | FID: 54.04 -> 33.94 | Improved |
| add image regularization to sketch loss (ablation) | L_sketch · Standing cat | FID: 50.09 -> 33.94 | Improved |
| add image regularization to sketch loss (ablation) | L_sketch · Horse on a side | FID: 40.65 -> 29.62 | Improved |
| add image regularization to sketch loss (ablation) | L_sketch · Horse rider | FID: 27.39 -> 27.50 | No clear effect |
| add weight regularization to sketch loss (ablation) | L_sketch · Standing cat | FID: 50.09 -> 49.76 | No clear effect |
| use pre-trained weights for the sketch discriminator (vs D scratch) (ablation) | Full (w/o aug.) · Horse on a side | FID: 27.84 -> 29.62 | Improved |
| tune whole mapping network instead of only W-shift bias (ablation) | Full (w/o aug.) · Horse rider | FID: 30.66 -> 27.50 | Improved |
| add translation augmentation (ablation) | Full · Horse rider | FID: 27.50 -> 19.94 | Improved |
| add translation augmentation (ablation) | Full · Gabled church | FID: 16.70 -> 21.35 | Hurt |
| train with 30 sketches instead of 1 (ablation) | Full · Horse rider | FID: 29.25 -> 27.50 | Improved |
| train with 30 sketches instead of 5 (ablation) | Full · Standing cat | FID: 31.20 -> 33.94 | Hurt |
Closest papers by meaning
- SketchyGAN: Towards Diverse and Realistic Sketch to Image Synthesis (CVPR 2018 · similarity 0.90)
- Image Generation from Sketch Constraint Using Contextual GAN (ECCV 2018 · similarity 0.89)
- SketchGAN: Joint Sketch Completion and Recognition With Generative Adversarial Network (CVPR 2019 · similarity 0.89)
- StudioGAN: A Taxonomy and Benchmark of GANs for Image Synthesis (TPAMI 2023 · similarity 0.87)
- Creative Sketch Generation (ICLR 2021 · similarity 0.87)
- A Large-Scale Study on Regularization and Normalization in GANs (ICML 2019 · 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.
