search

zhanglonghao1992 / One-Shot_Free-View_Neural_Talking_Head_Synthesis

Pytorch implementation of paper "One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing"
Other
735 stars 141 forks source link

Does quantitative evaluation results match the ones in the original paper? #36

Closed YifengMa9 closed 2 years ago

YifengMa9 commented 2 years ago

Thank you so much for your implementation. I wonder whether you have quantitatively evaluated your implementation and compare your results with ones in original paper.

Have you calculated L1, PNSR, MS-SSIM scores?

zhanglonghao1992 commented 2 years ago

@lostork To be honest, I don't care much about these scores. I think the generation algorithms based on GANs should be evaluated mainly through observation. Of course, except for publishing papers :)