Research map / Image generation / GANs
Geometry-Consistent Generative Adversarial Networks for One-Sided Unsupervised Domain Mapping
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 |
|---|---|---|---|
| add geometry consistency with rotation | GAN alone · Cityscapes image → parsing | class acc: 0.160 -> 0.234 | Helped |
| add geometry consistency with rotation | GAN alone · Cityscapes parsing → image | mean IoU: 0.098 -> 0.129 | Helped |
| add geometry consistency with rotation | GAN alone · SVHN → MNIST | class acc (%): 26.8 -> 32.5 | Helped |
| use vertical flipping instead of rotation (ablation) | GcGAN · SVHN → MNIST | class acc (%): 32.5 -> 33.3 | Helped |
| do not share generator parameters (ablation) | GcGAN-rot · Cityscapes image → parsing | mean IoU: 0.170 -> 0.170 | No clear effect |
| mix rotation and vertical flipping transformations (ablation) | GcGAN-rot · Cityscapes image → parsing | mean IoU: 0.170 -> 0.168 | Hurt |
| add cycle consistency (ablation) | GcGAN-rot · Cityscapes image → parsing | mean IoU: 0.170 -> 0.182 | Helped |
| add distance preservation (ablation) | GcGAN-rot · SVHN → MNIST | class acc (%): 32.5 -> 34.0 | Helped |
| add cycle consistency (ablation) | GcGAN-rot · SVHN → MNIST | class acc (%): 32.5 -> 33.8 | Helped |
| combine cycle consistency with distance preservation (ablation) | CycleGAN · SVHN → MNIST | class acc (%): 26.1 -> 18.0 | Hurt |
| mix rotation and vertical flipping transformations (ablation) | GcGAN-rot · Aerial photo → Map | RMSE: 28.31 -> 27.98 | Helped |
| do not share generator parameters (ablation) | GcGAN-rot · Aerial photo → Map | RMSE: 28.31 -> 30.25 | Hurt |
Closest papers by meaning
- A Geometric Analysis of Deep Generative Image Models and Its Applications (ICLR 2021 · similarity 0.90)
- Breaking the Cycle - Colleagues Are All You Need (CVPR 2020 · similarity 0.88)
- Unsupervised Pixel-Level Domain Adaptation With Generative Adversarial Networks (CVPR 2017 · similarity 0.88)
- Efficient Geometry-Aware 3D Generative Adversarial Networks (CVPR 2022 · similarity 0.88)
- On Translation and Reconstruction Guarantees of the Cycle-Consistent Generative Adversarial Networks (NeurIPS 2022 · similarity 0.88)
- What You See is What You GAN: Rendering Every Pixel for High-Fidelity Geometry in 3D GANs (CVPR 2024 · 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.