InvVis/dataloader.py at main · huayuan4396/InvVis · GitHub

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
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
from skimage import io
from PIL import Image
from random import randint
from torch.utils.data import Dataset
import numpy as np
import torch
from DHNutils import GetOption, SaveImageFromTensor


def to_rgb(image):
    rgb_image = Image.new("RGB", image.size)
    rgb_image.paste(image)
    return rgb_image


class DHNDataset(Dataset):
    def __init__(self, imgNameList, dataNameList, qrNameList, imgTransform=None, dataTransform=None, qrTransform=None, isTrain=True):
        self.imgNameList = imgNameList
        self.dataNameList = dataNameList
        self.dataTypeNum = len(dataNameList)
        self.qrNameList = qrNameList
        self.imgTransform = imgTransform
        self.dataTransform = dataTransform
        self.qrTransform = qrTransform
        self.isTrain = isTrain

    def __len__(self):
        return len(self.imgNameList)

    def __getitem__(self, idx):
        # img = io.imread(self.imgNameList[idx])[:, :, :3]
        # img2 = io.imread(self.imgNameList[np.random.randint(0, len(self.imgNameList))])[:, :, :3]
        img = Image.open(self.imgNameList[idx])
        img = to_rgb(img)
        typeIdx = idx % self.dataTypeNum

        randData = io.imread(self.dataNameList[typeIdx][np.random.randint(0, len(self.dataNameList[typeIdx]) - 1)])
        if len(randData.shape) == 2:
            pass
        elif len(randData.shape) == 3:
            randData = randData[:, :, 0]
        randData = Image.fromarray(randData)


        if self.isTrain:
            if self.imgTransform:
                img = self.imgTransform(img)
            if self.dataTransform:
                randData = self.dataTransform(randData)

            # randData = randData + torch.randn_like(randData) * 0.05

            randDataMin = torch.min(randData)
            randDataMax = torch.max(randData)
            randData = (randData - randDataMin) / (randDataMax - randDataMin + 0.00001)

            for i in range(GetOption("embedChannelNum") - 2):
                now = io.imread(self.dataNameList[typeIdx][np.random.randint(0, len(self.dataNameList[typeIdx]) - 1)])
                if len(now.shape) == 2:
                    pass
                elif len(now.shape) == 3:
                    now = now[:, :, 0]
                now = Image.fromarray(now)

                if self.dataTransform:
                    now = self.dataTransform(now)

                # now = now + torch.randn_like(now) * 0.05

                nowMin = torch.min(now)
                nowMax = torch.max(now)
                now = (now - nowMin) / (nowMax - nowMin + 0.00001)

                randData = torch.cat([randData, now], dim=0)

            qr = io.imread(self.qrNameList[np.random.randint(0, len(self.qrNameList) - 1)])
            if len(qr.shape) == 2:
                pass
            elif len(qr.shape) == 3:
                qr = qr[:, :, 0]
            qr = Image.fromarray(qr)

            if self.qrTransform:
                qr = self.qrTransform(qr)

            randData = torch.cat([randData, qr * GetOption("qrMul")], dim=0)


            return {"image": img, "randData": randData}

        else:
            randData = io.imread(self.dataNameList[typeIdx][idx % len(self.dataNameList[typeIdx])])
            if len(randData.shape) == 2:
                pass
            elif len(randData.shape) == 3:
                randData = randData[:, :, 0]
            randData = Image.fromarray(randData)

            if self.imgTransform:
                img = self.imgTransform(img)
            if self.dataTransform:
                randData = self.dataTransform(randData)

            # randData = randData + torch.randn_like(randData) * 0.05

            randDataMin = torch.min(randData)
            randDataMax = torch.max(randData)
            randData = (randData - randDataMin) / (randDataMax - randDataMin + 0.00001)

            for i in range(GetOption("embedChannelNum") - 2):
                now = io.imread(self.dataNameList[typeIdx][(idx + i + 1) % len(self.dataNameList[typeIdx])])
                if len(now.shape) == 2:
                    pass
                elif len(now.shape) == 3:
                    now = now[:, :, 0]
                now = Image.fromarray(now)

                if self.dataTransform:
                    now = self.dataTransform(now)

                # now = now + torch.randn_like(now) * 0.05

                nowMin = torch.min(now)
                nowMax = torch.max(now)
                now = (now - nowMin) / (nowMax - nowMin + 0.00001)

                randData = torch.cat([randData, now], dim=0)

            qr = io.imread(self.qrNameList[idx % len(self.qrNameList)])
            if len(qr.shape) == 2:
                pass
            elif len(qr.shape) == 3:
                qr = qr[:, :, 0]
            qr = Image.fromarray(qr)

            if self.qrTransform:
                qr = self.qrTransform(qr)

            return {"image": img, "dataImg": randData, "qrImg": qr * GetOption("qrMul")}