You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First of all thank you for this work it is wonderful.
I need help, I want to train my own model but I have this questions.
In the documentation of the dataset I see there are three categories but I dont know which categories is each one, I Suppose that 1 is Real and 0 is Spoofing but why is there a third category.
To train the model we need to resize the images as it indicates?, or there's a way to do it automatically maybe a script?
There's a way to use your model as a pretrained model.
Thank you for you help
The text was updated successfully, but these errors were encountered:
First of all thank you for this work it is wonderful.
I need help, I want to train my own model but I have this questions.
Thank you for you help
The text was updated successfully, but these errors were encountered: