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
782 stars 143 forks source link

Could you share the vox dataset? #63

Open brianw0924 opened 2 years ago

brianw0924 commented 2 years ago

The official website removed the dataset link

could you provide the vox dataset?

zhanglonghao1992 commented 2 years ago

@brianw0924 https://github.com/AliaksandrSiarohin/video-preprocessing

brianw0924 commented 2 years ago

I know this repo but it takes really really long time to run the script (download from youtube + crop) without 8 V100 GPU.

I only have 1 3080, it takes like >100 days to finish the script (according to tqdm)

Could you kindly provide the processed video link?

yahskapar commented 2 years ago

@brianw0924, were you able to get the processed video link somehow? I see there's no response from @zhanglonghao1992 yet, unfortunately. I'm running into the same issue you did with the previously linked video pre-processing repo requiring quite a bit of computation to utilize.

sky24h commented 1 year ago

@brianw0924 The crop should not consume GPU resources. It sounds more like a network problem if you need >100 days.