CycleGANN IMDB to Rijksmuseum full painting
CycleGANN IMDB to Rijksmuseum face only (this file)
Perceived sex classification performance
Source: https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
Cite:
@article{zhu2017unpaired,
title={Unpaired image-to-image translation using cycle-consistent adversarial networks},
author={Zhu, Jun-Yan and Park, Taesung and Isola, Phillip and Efros, Alexei A},
journal={arXiv preprint},
year={2017}
}Author's implementation on the Monet2Photo model is trained with 1072 Monet painting samples and 6,287 nature photographs over 200 epochs (takes approximately 15 days with a single GPU). We have used 1234 Rijksmuseum paintings and 10,223 IMDb pictures with the same architecture. Training process took almost 18 days on Intel i5-6600 @3.30GHz equipped with NVIDIA GeForge GTX 1070 (Aero 8G OC V330 – 011R).
Faces from Rijksmuseum paintings → Painting face to photo results
→ 
→ 
→ 
→ 
→ 
→ 
→ 
→ 
→ 