search

PhilippeMarcotte / Deep-HDR-with-Pytorch

14 stars 6 forks source link

Why are there two datasets? #1

Open Feiyue-Peng opened 6 years ago

Feiyue-Peng commented 6 years ago

I notice that you used two datasets when training, one is training_data folder, the other is scene folder. I don't understand the reason. Could you please explain for me?

PhilippeMarcotte commented 6 years ago

Scene folder has one folder for each scene containing the three bracketed images and its HDR image. These are used to compute the training set.

The training_data folder contains the training set resulting from the execution of PrepareTrainingData.py on the scene folder.

I do not provide the training set directly because the folder takes up a lot of memory, about 80-90GB.