search

imlixinyang / HiSD

Code for "Image-to-image Translation via Hierarchical Style Disentanglement" (CVPR 2021 Oral).
Other
391 stars 49 forks source link

Data imbalance #29

Open ryanqiutu opened 2 years ago

ryanqiutu commented 2 years ago

Dear Author, I appreciate your work, but i have some questions about the data imbalance. The face number with glasses is only 1377 as the numbe without glasses is 25622. I want to known the effects of the data imbalance. Thanks~

imlixinyang commented 2 years ago

Thanks for your attention of HiSD. The imbalance problem can be widely seen in almost all unpaired datasets in I2I area. In most cases, it would be better if you have balanced dataset. Our dataset is directly converted by the label annotation file of CelebA. Since the labels are unbalanced, the numbers of reorganized images are unbalanced, too. However, if you want to focus on this, the results may be improved if you apply some tricks in long-tailed or unbalanced-labels area.