Imaginaire
October 13, 2021 ยท View on GitHub
Imaginaire
Docs | License | Installation | Model Zoo
Model Zoo
Introduction
We provide a wide range of pretrained imaginaire models for different tasks. All the models were trained using an NVIDIA DGX 1 machine with 8 32GB V100 using NVIDIA PyTorch docker 20.06 or later.
| Algorithm Name | Task | Model | Resolution |
|---|---|---|---|
| pix2pixHD | Cityscapes, segmentation to image | download | 1024x512 |
| SPADE | COCO-Stuff, segmentation to image | download | 256x256 |
Unsupervised Image-to-Image Translation
| Algorithm Name | Task | Model | Resolution |
|---|---|---|---|
| UNIT | Winter <-> Summer | download | 256x256 |
| MUNIT | AFHQ Dog <-> Cat | download | 256x256 |
| FUNIT | AniamlFaces | download | 256x256 |
| COCO-FUNIT | AniamlFaces | download | 256x256 |
| COCO-FUNIT | Mammals, Full body animal translation | download | 256x256 |
Video-to-video Translation
| Algorithm Name | Task | Model | Resolution |
|---|---|---|---|
| vid2vid | Cityscapes, segmentation to video | download | 1024x512 |
| fs-vid2vid | FaceForensics, landmarks to video | download | 512x512 |
World-to-world Translation
| Algorithm Name | Task | Model | Resolution |
|---|---|---|---|
| wc-vid2vid | Cityscapes, segmentation to video | download | 1024x512 |
| GANcraft | Semantic block worlds to realistic worlds | download | arbitrary |