search

amir-abdi / disentanglement-pytorch

Disentanglement library for PyTorch
GNU General Public License v3.0
273 stars 30 forks source link

Custom dataset #89

Open DianeBouchacourt opened 3 years ago

DianeBouchacourt commented 3 years ago

Hi,

Thank you very much for setting this library ! If I am correct, one can use a custom dataset by creating the corresponding data_loader, and updating accordingly the function:

https://github.com/amir-abdi/disentanglement-pytorch/blob/2626252148a0130fc58e9d97397b001930bf672b/common/data_loader.py#L182

This will run with the training and evaluation metrics code? There is nothing specific to dSprites and CelebA?

amir-abdi commented 3 years ago

Hi, I don't think I have made much assumptions on dSprites and/or CelebA datasets. However, looking at the code, I hope I had implemented the data loader following a more object oriented design so that you can create your own dataloader for your own dataset without having to deal with the if/else scenarios for dSprites and CelebA.

Anyhow, to answer your question: yes, you can add your own dataset following what I have done for celebA and dSprites.