Image-to-Image Translation
Image-to-Image Translation is a task in computer vision and machine learning where the goal is to learn a mapping between an input image and an output image, such that the output image can be used to perform a specific task, such as style transfer, data augmentation, or image restoration.
( Image credit: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks )
Papers
Showing 1–10 of 1184 papers
All datasetsSYNTHIA-to-CityscapesGTAV-to-Cityscapes LabelsCityscapes Labels-to-PhotoADE20K Labels-to-PhotosCOCO-Stuff Labels-to-PhotosADE20K-Outdoor Labels-to-PhotosCelebA-HQCityscapes-to-Foggy Cityscapescat2dogCityscapes Photo-to-LabelsBCIFLIR
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Pix2PixHD-AUG | Accuracy | 77.9 | — | Unverified |
| 2 | DP-SIMS (ConvNext-L) | mIoU | 54.3 | — | Unverified |
| 3 | DP-GAN | mIoU | 52.7 | — | Unverified |
| 4 | SCDM | mIoU | 49.4 | — | Unverified |
| 5 | SPADE + SESAME | mIoU | 49 | — | Unverified |
| 6 | OASIS | mIoU | 48.8 | — | Unverified |
| 7 | CC-FPSE-AUG | mIoU | 44 | — | Unverified |
| 8 | CC-FPSE | mIoU | 43.7 | — | Unverified |
| 9 | SPADE | mIoU | 38.5 | — | Unverified |
| 10 | VQGAN+Transformer | FID | 35.5 | — | Unverified |