L1-SR-GUI/SRCNN-pytorch-master/datasets.py
2020-11-12 23:30:33 +01:00

31 lines
934 B
Python

import h5py
import numpy as np
from torch.utils.data import Dataset
class TrainDataset(Dataset):
def __init__(self, h5_file):
super(TrainDataset, self).__init__()
self.h5_file = h5_file
def __getitem__(self, idx):
with h5py.File(self.h5_file, 'r') as f:
return np.expand_dims(f['lr'][idx] / 255., 0), np.expand_dims(f['hr'][idx] / 255., 0)
def __len__(self):
with h5py.File(self.h5_file, 'r') as f:
return len(f['lr'])
class EvalDataset(Dataset):
def __init__(self, h5_file):
super(EvalDataset, self).__init__()
self.h5_file = h5_file
def __getitem__(self, idx):
with h5py.File(self.h5_file, 'r') as f:
return np.expand_dims(f['lr'][str(idx)][:, :] / 255., 0), np.expand_dims(f['hr'][str(idx)][:, :] / 255., 0)
def __len__(self):
with h5py.File(self.h5_file, 'r') as f:
return len(f['lr'])