UMRCH/dataset.py at main · Lab201A/UMRCH · 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
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
from torchvision import transforms
from PIL import Image
import numpy as np
import random
from torch.utils.data import Dataset
import os
import json

WORKDIR = ''
train_transform = transforms.Compose([
    transforms.RandomHorizontalFlip(),
    transforms.Resize(224),
    transforms.RandomCrop(224),
    transforms.ToTensor(),
    transforms.Normalize((0.48145466, 0.4578275, 0.40821073), (0.26862954, 0.26130258, 0.27577711)),
])
test_transform = transforms.Compose([
    transforms.Resize(224),
    transforms.RandomCrop(224),
    transforms.ToTensor(),
    transforms.Normalize((0.48145466, 0.4578275, 0.40821073), (0.26862954, 0.26130258, 0.27577711)),
])

class BaseHashDataset(Dataset):
    def __init__(self, label_file, img_root, img_transform, split_list, split, sample_size, contrastive=False) -> None:
        super().__init__()
        self.img_root = img_root
        self.img_transform =img_transform
        self.contrastive = contrastive
        img_labels = open(label_file, 'r').readlines()
        self.img_label_list = [(
            os.path.join(img_root, val.split()[0]),
            np.array([int(la) for la in val.split()[1:]])
            ) for val in img_labels]

        if sample_size is None:
            self.global_index = np.array(split_list[split])
        else:
            _size = sum([len(split_list[key]) for key in ['query', 'db']])
            sample_index = random.sample(list(range(_size)), sample_size)
            self.global_index = np.array(sample_index)

    def get_image(self, imgpath):
        with open(imgpath, 'rb') as imgf:
            img = Image.open(imgf).convert('RGB')
        return img
    def get_image(self, imgpath):
        try:
            # img = Image.open(img_path).convert('RGB')
            with open(imgpath, 'rb') as imgf:
                img = Image.open(imgf).convert('RGB')
            return img
        except Exception as e:
            print(f"bad: {imgpath}")
            raise e

    def __getitem__(self, index):
        img_path, label = self.img_label_list[self.global_index[index]]
        image = self.get_image(img_path)
        if self.contrastive and (self.img_transform == train_transform) :
            image1 = self.img_transform(image)
            image2 = self.img_transform(image)
            return index, image1, label, image2
        elif self.img_transform is not None:
            image = self.img_transform(image)
            return index, image, label

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

class MIRFlickrHashDataset(BaseHashDataset):
    def __init__(
            self,
            img_transform, split,
            sample_size = None,
            label_file = './data/flickr25k/allannots.txt',
            img_root = 'dir to mirflickr',
            contrastive = False
        ):
        assert split in ['train', 'db', 'query']
        if os.path.exists('./data/hash_split_for_mirflickr.json'):
            split_list = json.load(open('./data/hash_split_for_mirflickr.json', 'r'))
        else:
            all_index = list(range(len(open(label_file, 'r').readlines())))
            query_index = random.sample(all_index, 2000)
            train_index = random.sample(list(set(all_index)-set(query_index)), 5000)
            db_index = list(set(all_index)-set(query_index))
            split_list = {
                'train': train_index,
                'db': db_index,
                'query': query_index
            }
            json.dump(split_list, open('./data/hash_split_for_mirflickr.json', 'w'))
        super().__init__(label_file, img_root, img_transform, split_list, split, sample_size, contrastive)

class COCO14HashDataset(BaseHashDataset):
    def __init__(
            self,
            img_transform, split,
            sample_size = None,
            label_file = './data/coco2014/allannots.txt',
            img_root = 'dir to coco2014',
            contrastive = False
        ):
        assert split in ['train', 'db', 'query']
        if os.path.exists('./data/hash_split_for_coco14.json'):
            split_list = json.load(open('./data/hash_split_for_coco14.json', 'r'))
        else:
            all_index = list(range(len(open(label_file, 'r').readlines())))
            query_index = random.sample(all_index, 5000)
            train_index = random.sample(list(set(all_index)-set(query_index)), 10000)
            db_index = list(set(all_index)-set(query_index))
            split_list = {
                'train': train_index,
                'db': db_index,
                'query': query_index
            }
            json.dump(split_list, open('./data/hash_split_for_coco14.json', 'w'))
        super().__init__(label_file, img_root, img_transform, split_list, split, sample_size, contrastive)

class NUSWideHashDataset1(Dataset):
    def __init__(
            self,
            img_transform,
            split,
            sample_size=None,
            img_root='dir to nuswide',
            contrastive=False
        ):
        assert split in ['train', 'db', 'query'], "split must be one of ['train', 'db', 'query']"

        train_file = './data/nuswide/train.txt'
        db_file = './data/nuswide/database.txt'
        query_file = './data/nuswide/test.txt'

        self.img_transform = img_transform
        self.img_root = img_root
        self.contrastive = contrastive

        if split == 'train':
            self.split_file = train_file
        elif split == 'db':
            self.split_file = db_file
        elif split == 'query':
            self.split_file = query_file

        self.img_label_list = self._load_split(self.split_file)

        self.global_index = np.array(range(len(self.img_label_list)))
        if sample_size is not None:
            sample_index = random.sample(range(len(self.img_label_list)), sample_size)
            self.global_index = np.array(sample_index)

    def _load_split(self, split_file):
        with open(split_file, 'r') as f:
            lines = f.readlines()

        img_label_list = []
        for line in lines:
            parts = line.strip().split()
            img_path = os.path.join(self.img_root, parts[0])
            label = np.array([int(la) for la in parts[1:]])
            img_label_list.append((img_path, label))

        return img_label_list

    def get_image(self, imgpath):
        img_path = os.path.join(self.img_root, imgpath)
        with open(img_path, 'rb') as imgf:
            img = Image.open(imgf).convert('RGB')
        return img

    def __getitem__(self, index):
        img_path, label = self.img_label_list[self.global_index[index]]
        image = self.get_image(img_path)

        if self.contrastive and (self.img_transform == train_transform):
            image1 = self.img_transform(image)
            image2 = self.img_transform(image)
            return index, image1, label, image2
        elif self.img_transform is not None:
            image = self.img_transform(image)
            return index, image, label

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