- Download dataset from here
- Extract 50,000 face from dataset using "dataset_preparer.py"
- [optional] Plot your dataset using "plot_dataset.py" file
- [optional] If you cant run step 5 because of internet noises, use "partitioner.py" to covert the "celeba.npz" to 5 part and mix them all throw "train.py"
- Upload the "celeba.npz" file on your google drive
helli5-dev / fakeface Goto Github PK
View Code? Open in Web Editor NEWLicense: GNU General Public License v3.0