Statements (51)
Predicate | Object |
---|---|
gptkbp:instanceOf |
gptkb:convolutional_neural_network
|
gptkbp:application |
image synthesis
face generation artificial image generation |
gptkbp:architecture |
gptkb:GAN
|
gptkbp:author |
gptkb:Jaakko_Lehtinen
gptkb:Samuli_Laine gptkb:Tero_Karras gptkb:Timo_Aila Jan Karras Miika Aittala |
gptkbp:citation |
2020
over 5000 |
gptkbp:developedBy |
gptkb:NVIDIA
|
gptkbp:field |
gptkb:artificial_intelligence
gptkb:machine_learning computer vision |
https://www.w3.org/2000/01/rdf-schema#label |
StyleGAN2
|
gptkbp:improves |
better FID scores
better perceptual quality improved path length regularization removal of blob artifacts revised discriminator architecture revised generator architecture weight demodulation |
gptkbp:input |
latent vector
|
gptkbp:license |
gptkb:NVIDIA_Source_Code_License
|
gptkbp:notableFeature |
improved image quality
adaptive discriminator augmentation reduced artifacts |
gptkbp:notableFor |
AI-generated portraits
style mixing artificial face datasets latent space exploration synthetic datasets |
gptkbp:notableImprovementOver |
gptkb:StyleGAN
|
gptkbp:notablePublication |
Analyzing and Improving the Image Quality of StyleGAN
|
gptkbp:output |
synthetic image
|
gptkbp:predecessor |
gptkb:StyleGAN
|
gptkbp:publishedIn |
Conference on Computer Vision and Pattern Recognition (CVPR)
|
gptkbp:releaseYear |
2019
|
gptkbp:repository |
https://github.com/NVlabs/stylegan2
|
gptkbp:successor |
gptkb:StyleGAN3
|
gptkbp:trainer |
gptkb:CelebA-HQ
gptkb:FFHQ gptkb:LSUN |
gptkbp:usedFor |
data augmentation
art creation deepfake generation |
gptkbp:bfsParent |
gptkb:StyleGAN
|
gptkbp:bfsLayer |
6
|