KD-AFA-Net/dataset.py at main · junseo0421/KD-AFA-Net · 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
from torch.utils.data import Dataset
from PIL import Image
import numpy as np
import random
from os.path import join, splitext, basename

def rand_crop(img, target_height, target_width):
    # reshape image to an appropriate size, and random crop to target size

    width = img.size[0]
    height = img.size[1]

    width_scale = target_width / width
    height_scale = target_height / height

    if height_scale >= 0.5:
        starting_x = random.randint(0, width - target_width)
        starting_y = random.randint(0, height - target_height)
    else:
        down_sample_ratio = height_scale / 0.5
        if round(down_sample_ratio*width) < target_width:
            down_sample_ratio = width_scale
        new_width = round(down_sample_ratio * width)
        new_height = round(down_sample_ratio * height)
        img = img.resize((new_width, new_height))
        starting_x = random.randint(0, new_width - target_width)
        starting_y = random.randint(0, new_height - target_height)

    img = img.crop((starting_x, starting_y, starting_x+target_width, starting_y+target_height))

    return img


def center_crop(img, target_height, target_width):
    # reshape image to an appropriate size, and center crop to target size

    width = img.size[0]
    height = img.size[1]

    width_scale = target_width / width
    height_scale = target_height / height

    if height_scale >= 0.5:
        starting_x = (width - target_width) / 2
        starting_y = (height - target_height) / 2
    else:
        down_sample_ratio = height_scale / 0.5
        if round(down_sample_ratio*width) < target_width:
            down_sample_ratio = width_scale
        new_width = round(down_sample_ratio * width)
        new_height = round(down_sample_ratio * height)
        img = img.resize((new_width, new_height))
        starting_x = (new_width - target_width) / 2
        starting_y = (new_height - target_height) / 2

    img = img.crop((starting_x, starting_y, starting_x+target_width, starting_y+target_height))

    return img

class dataset_norm(Dataset):
    def __init__(self, root='', transforms=None, imgSize=192, inputsize=128, imglist1=[], imglist2=[], imglist3=[]):
        self.transforms = transforms
        self.imgSize = imgSize
        self.inputsize = inputsize

        self.img_list1 = imglist1
        self.img_list2 = imglist2
        self.img_list3 = imglist3

        self.size = len(self.img_list1)

    def __getitem__(self, index):
        index = index % self.size

        img1 = Image.open(self.img_list1[index]).convert("L")
        img2 = Image.open(self.img_list2[index]).convert("L")
        img3 = Image.open(self.img_list3[index]).convert("L")

        img1 = np.array(img1)
        img2 = np.array(img2)
        img3 = np.array(img3)

        img_cat = np.stack([img1, img2, img3], axis=-1)
        img_cat = Image.fromarray(img_cat)
        img_cat = self.transforms(img_cat)

        c0 = img_cat[0].unsqueeze(0)
        gt = c0.repeat(3, 1, 1)

        i = (self.imgSize - self.inputsize) // 2

        iner_img = img_cat[:, :, i:i + self.inputsize]
        mask_img = np.ones((3, self.imgSize, self.imgSize))
        mask_img[:, :, i:i + self.inputsize] = iner_img

        return gt, mask_img

    def __len__(self):
        return self.size


class dataset_test4(Dataset):
    def __init__(self, root='', transforms=None, imgSize=192, inputsize=128, pred_step=1, imglist=[]):
        self.pred_step = pred_step
        self.transforms = transforms
        self.imgSize = imgSize
        self.preSize = imgSize + 64 * (pred_step - 1)
        self.inputsize = inputsize
        self.inputsize2 = inputsize + 64 * (pred_step - 1)

        self.img_list = imglist
        self.size = len(self.img_list)

    def __getitem__(self, index):
        index = index % self.size
        name = self.img_list[index]
        img = Image.open(name).convert('RGB')

        i = (self.imgSize - self.inputsize) // 2

        if self.transforms is not None:
            img = self.transforms(img)

        iner_img = img

        mask_img = np.ones((3, self.preSize, self.preSize))

        if self.pred_step > 1:
            mask_img[:, i:i + self.inputsize2, i:i+self.inputsize2] = img
        else:
            mask_img[:, :, i:i + self.inputsize2] = iner_img

        return img, iner_img, mask_img, splitext(basename(name))[0], name.replace('\\', '/').split('/')[-2]

    def __len__(self):
        return self.size