-
Notifications
You must be signed in to change notification settings - Fork 0
/
data_prep.py
46 lines (38 loc) · 1.41 KB
/
data_prep.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
import pickle
import numpy as np
import torch
from torch.utils.data import Dataset
from PIL import Image
class ImageDataset(Dataset):
"""MIT image dataset."""
def __init__(self, root_dir, data, transform = None):
self.root_dir = root_dir
self.dataPtr = data
self.transform = transform
with open(self.root_dir + "cleanTrainCap", "rb") as f:
self.trainCap = pickle.load(f)
with open("cleanTrainID", "rb") as f:
self.trainID = pickle.load(f) - 1
def __len__(self):
return len(self.dataPtr)
def __getitem__(self, idx):
img_name = self.root_dir + "images/" + self.dataPtr[idx]
image = Image.open(img_name)
if image.mode == 'L':
image = image.convert('RGB')
elif image.mode == 'CMYK':
image = image.convert('RGB')
# if image.size[0] == 1:
# image = image.repeat(3,1,1)
#
# elif image.size[0] == 4:
# image = image[0].repeat(3,1,1)
if self.transform:
image = self.transform(image)
imgNum = int(self.dataPtr[idx][3:-4])
# captions = self.trainCap[self.trainID == imgNum]
# rand = np.random.randint(captions.shape[0])
# caption = captions[rand]
# sample = {'image': image, 'caption': torch.from_numpy(np.array(caption))}
sample = image
return sample, imgNum