Research map / Image generation / GANs
Self-Supervised Dense Consistency Regularization for Image-to-Image Translation
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 DCR to CycleGAN | CycleGAN · Horse →Zebra | FID: 78.2±1.0 -> 54.4±3.3 | Helped |
| add DCR to CUT | CUT · Horse →Zebra | FID: 43.2±2.3 -> 34.0±0.4 | Helped |
| add DCR to MUNIT | MUNIT · Horse →Zebra | FID: 102.3±4.3 -> 59.9±0.1 | Helped |
| add DCR to CUT | CUT · Winter →Summer | FID: 77.8±0.5 -> 73.2±0.5 | Helped |
| add DCR to StarGANv2 | StarGANv2 · Cat →Dog (AFHQ) | FID: 44.2 -> 33.0 | Helped |
| add DCR to StarGANv2 | StarGANv2 · AFHQ wild domain, 10% training data | FID: 22.63 ± 3.70 -> 17.15 ± 0.89 | Helped |
| add DCR to SNDCGAN | SNDCGAN · CIFAR-10 | FID: 97.4 -> 8.6 | Helped |
| add DCR alongside ContraD | SNDCGAN + ContraD · CIFAR-10 | FID: 10.9 -> 7.7 | Helped |
| apply DCR at layer2 (ablation) | CycleGAN · Horse →Zebra | FID: 77.2 -> 49.2 | Helped |
| apply DCR at layer3 instead of layer2 (ablation) | CycleGAN · Horse →Zebra | FID: 49.2 -> 51.4 | Hurt |
| apply DCR at layer4 instead of layer2 (ablation) | CycleGAN · Horse →Zebra | FID: 49.2 -> 51.1 | Hurt |
| apply DCR at layer4 (ablation) | CycleGAN · Cat →Dog (ImageNet) | FID: 86.5 -> 57.8 | Helped |
Closest papers by meaning
- TraVeLGAN: Image-To-Image Translation by Transformation Vector Learning (CVPR 2019 · similarity 0.91)
- NICE: NoIse-modulated Consistency rEgularization for Data-Efficient GANs (NeurIPS 2023 · similarity 0.90)
- Improved Consistency Regularization for GANs (AAAI 2021 · similarity 0.90)
- Compressing Image-to-Image Translation GANs Using Local Density Structures on Their Learned Manifold (AAAI 2024 · similarity 0.89)
- Rethinking the Paradigm of Content Constraints in Unpaired Image-to-Image Translation (AAAI 2024 · similarity 0.87)
- Style Quantization for Data-Efficient GAN Training (CVPR 2025 · 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.