-
Notifications
You must be signed in to change notification settings - Fork 0
/
VOCSegDataset.py
68 lines (51 loc) · 2.43 KB
/
VOCSegDataset.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
import random
import torch
import torchvision.transforms as transforms
from torch.utils.data import DataLoader
from torchvision.datasets import VOCSegmentation
import numpy as np
from PIL import Image
class VOCSegDataset(VOCSegmentation):
def __getitem__(self, idx):
image = Image.open(self.images[idx]).convert('RGB')
label = Image.open(self.masks[idx])
if self.transforms is not None:
seed = random.randint(0, 2 ** 32)
self._set_seed(seed); image = self.transforms(image)
self._set_seed(seed); label = self.transforms(label) * 255
label[label > 20] = 0
return image, label
def _set_seed(self, seed):
random.seed(seed)
torch.manual_seed(seed)
if __name__=='__main__':
image_transforms = transforms.Compose([
transforms.Resize((256, 256)),
transforms.RandomHorizontalFlip(p=0.5),
transforms.RandomRotation((-5, 5), expand=False),
transforms.ToTensor(),
])
target_transforms = transforms.Compose([
transforms.Resize((256, 256)),
transforms.RandomHorizontalFlip(p=0.5),
transforms.RandomRotation((-5, 5), expand=False),
transforms.ToTensor(),
])
train_ds = VOCSegDataset('/media/jhnam0514/68334fe0-2b83-45d6-98e3-76904bf08127/home/namjuhyeon/Desktop/LAB/AwesomeDeepLearning/dataset/IS2D_dataset/PASCAL VOC/',
year='2012', image_set='train', download=True, transforms=image_transforms)
val_ds = VOCSegDataset('/media/jhnam0514/68334fe0-2b83-45d6-98e3-76904bf08127/home/namjuhyeon/Desktop/LAB/AwesomeDeepLearning/dataset/IS2D_dataset/PASCAL VOC/',
year='2012', image_set='val', download=True, transforms=image_transforms)
np.random.seed(0)
num_classes = 21
COLORS = np.random.randint(0, 2, size=(num_classes + 1, 3), dtype='uint8')
train_loader = DataLoader(train_ds, batch_size=2, shuffle=True)
import matplotlib.pyplot as plt
for img, target in train_loader:
fig, ax = plt.subplots(1, 2)
ax[0].imshow(np.transpose(img[0].cpu().detach().numpy(), (1, 2, 0)))
ax[0].axis('off'); ax[0].set_xticks([]); ax[0].set_yticks([])
ax[1].imshow(target[0].squeeze().cpu().detach().numpy())
ax[1].axis('off'); ax[1].set_xticks([]); ax[1].set_yticks([])
plt.tight_layout()
plt.subplots_adjust(left=0, bottom=0, right=1, top=1, hspace=0, wspace=0)
plt.show()