Closed jimydavis closed 3 months ago
me too
Thank you
Hi, we released the three datasets we collected in this work. Previously released datasets are not included.
To get id_template for BIWI or Vocaset, you can download the datasets following the link in README, and then store their template.pkl file into id_template.npy by throw the key and save just the vertices.
Note that, the id order shold be consistent with train.json. Please download our dataset to see the data structure.
Hi is this file supposed to be inside one of the downloads?
No such file or directory: './unitalker_data_release/BIWI/id_template.npy'
Thank you
Hi, we just uploaded the id_template.zip to google cloud https://drive.google.com/file/d/1Xr3mHyIBf5Hdy8l16JOlnv0QMNFLp268/view?usp=sharing .
We will update README
Thank you very much! The original BIWI dataset is no longer available btw. Do you know if it moved to a different platform or something?
Thank you again.
Thank you very much! The original BIWI dataset is no longer available btw. Do you know if it moved to a different platform or something?
Thank you again.
I just found that too. I don't know how to get the original data in this case. We have the prepared data, but we can't distribute it due to the license issue.
If your target is to train the model, we recomend to use our released data first. Besides, you can you use meshtalk and vocaset, and 3DETF datasets. They have good data quality.
Hi is this file supposed to be inside one of the downloads?
No such file or directory: './unitalker_data_release/BIWI/id_template.npy'
Thank you