search

deepmo24 / CPEM

PyTorch implementation of "Towards Accurate Facial Motion Retargeting with Identity-Consistent and Expression-Exclusive Constraints" (AAAI2022)
95 stars 18 forks source link

Which is the test split of FEAFA dataset? #10

Closed gerwang closed 1 year ago

gerwang commented 2 years ago

Hi, In the released code, it seems that all the FEAFA dataset is used for training. I wonder which test split was used to generate Tab. 1 in the paper? Thank you!

deepmo24 commented 2 years ago

Hi, we divide 123 FEAFA videos into two parts, the last 103 videos as the training set and the first 20 videos as the test set, which is described in the supplementary file.

gerwang commented 2 years ago

Thank you! I will try to obtain the peak expression frames using threshold. I wonder if it is possible to make the data_files (npy) of feafa public?