Research map / Image generation / Editing and style transfer
Harmonic Unpaired Image-to-image Translation
Rui Zhang, Tomas Pfister, Jia Li
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 graph smoothness regularization using patch histogram features | CycleGAN on BRATS Flair → T1 · BRATS Flair → T1 | MSE: 674.40 -> 216.83 | Improved |
| Add graph smoothness regularization using patch VGG features | CycleGAN on BRATS T1 → Flair · BRATS T1 → Flair | MSE: 1026.19 -> 127.84 | Improved |
| Use VGG rather than histogram features for graph similarity | HarmonicGAN on Cityscapes · Cityscapes Label → Photo | Pixel accuracy: 52.2 -> 55.9 | Improved |
| Use histogram features for graph similarity rather than CycleGAN's baseline objective | CycleGAN on Cityscapes Label → Photo · Cityscapes Label → Photo | Pixel accuracy: 52.7 -> 52.2 | Hurt |
| Use VGG rather than histogram features for graph similarity | HarmonicGAN on BRATS Flair → T1 · BRATS Flair → T1 | MAE: 6.38 -> 6.86 | Improved |
| Use VGG rather than histogram features for graph similarity | HarmonicGAN on BRATS T1 → Flair · BRATS T1 → Flair | MAE: 5.04 -> 4.69 | Hurt |
| Add graph smoothness regularization with VGG features | CycleGAN on Cityscapes · Cityscapes Photo → Label | Pixel accuracy: 57.2 -> 59.8 | Improved |
| Add graph smoothness regularization | CycleGAN on BRATS · BRATS radiologist study | Prefer (%): 5 -> 95 | Improved |
| Add graph smoothness regularization | CycleGAN on horse → zebra · Horse → zebra user study | Mean Likert: 3.16 -> 3.60 | Improved |
| Add CRF pairwise regularization | CycleGAN on BRATS Flair → T1 · BRATS Flair → T1 | MSE: 674.40 -> 839.47 | Hurt |
| Add CRF pairwise regularization | CycleGAN on BRATS T1 → Flair · BRATS T1 → Flair | MSE: 1026.19 -> 1138.42 | Hurt |
Closest papers by meaning
- TraVeLGAN: Image-To-Image Translation by Transformation Vector Learning (CVPR 2019 · similarity 0.87)
- Unpaired Image-To-Image Translation Using Cycle-Consistent Adversarial Networks (ICCV 2017 · similarity 0.87)
- Breaking the Dilemma of Medical Image-to-image Translation (NeurIPS 2021 · similarity 0.87)
- Harmonic Networks: Deep Translation and Rotation Equivariance (CVPR 2017 · similarity 0.86)
- Semantically Robust Unpaired Image Translation for Data With Unmatched Semantics Statistics (ICCV 2021 · similarity 0.86)
- Homomorphic Interpolation Network for Unpaired Image-to-image Translation (TPAMI 2020 · similarity 0.86)
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.
