-
Notifications
You must be signed in to change notification settings - Fork 0
/
newLoader.py
68 lines (60 loc) · 2.32 KB
/
newLoader.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
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
# -- coding: utf-8 --
"""
Created on Sat Jan 11 13:13:24 2020
@author: basit
"""
import pickle
import numpy as np
import torch
from torch.utils.data import Dataset
from PIL import Image
import random
class ImageDataset(Dataset):
"""MIT image dataset."""
def __init__(self, root_dir, trainPercent, valPercent, flag, transform=None):
self.root_dir = root_dir
# self.dataPtr = sorted(os.listdir(root+"images/"), key=len)
self.transform = transform
# self.flag = flag
with open(self.root_dir + "cleanTrainCap", "rb") as f:
trainCaps = pickle.load(f)
with open("cleanTrainID", "rb") as f:
trainID = pickle.load(f) - 1
random.seed(35)
randInd = np.arange(len(trainID))
random.shuffle(randInd)
trainCaps = trainCaps[randInd]
trainID = trainID[randInd]
# np.random.shuffle(trainID)
lengths = [int(len(trainID)*trainPercent), int(len(trainID)*(trainPercent+valPercent))]#, (len(self.trainID) - (int(len(self.trainID)*trainPercent)+ int(len(self.trainID)*valPercent)))]
# train = trainID[:lengths[0]]
# val = trainID[lengths[0]:lengths[1]]
# test = trainID[lengths[1]:]
if flag == 'train':
self.imgID = trainID[:lengths[0]]
self.captions = trainCaps[:lengths[0]]
elif flag == 'validation':
self.imgID = trainID[lengths[0]:lengths[1]]
self.captions = trainCaps[lengths[0]:lengths[1]]
elif flag == 'test':
self.imgID = trainID[lengths[1]:]
self.captions = trainCaps[lengths[1]:]
def __len__(self):
return len(self.imgID)
def __getitem__(self, idx):
imgName = self.imgID[idx]
img_name = self.root_dir + "images/img" + str(imgName) + ".jpg"
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)
caption = self.captions[idx]
sample = {'image': image, 'caption': torch.from_numpy(np.array(caption))}
return sample