-
Notifications
You must be signed in to change notification settings - Fork 22
/
imagenet.py
283 lines (225 loc) · 10.3 KB
/
imagenet.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
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
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
import json
import os
import os.path
import h5py
import numpy as np
import torch
import torch.utils.data as data
import torchvision.datasets as datasets
import torchvision.transforms as transforms
from PIL import ImageEnhance
import bf3s.utils as utils
# Set the appropriate paths of the datasets here.
_IMAGENET_DATASET_DIR = "./datasets/ImageNet"
_IMAGENET256_DATASET_DIR = "/datasets_local/ImageNet256"
_IMAGENET_LOWSHOT_BENCHMARK_CATEGORY_SPLITS_PATH = (
"./data/IMAGENET_LOWSHOT_BENCHMARK_CATEGORY_SPLITS.json"
)
_MEAN_PIXEL = [0.485, 0.456, 0.406]
_STD_PIXEL = [0.229, 0.224, 0.225]
def load_ImageNet_fewshot_split(class_names):
with open(_IMAGENET_LOWSHOT_BENCHMARK_CATEGORY_SPLITS_PATH, "r") as f:
label_idx = json.load(f)
assert len(label_idx["label_names"]) == len(class_names)
def get_class_indices(class_indices1):
class_indices2 = []
for index in class_indices1:
class_name_this = label_idx["label_names"][index]
assert class_name_this in class_names
class_indices2.append(class_names.index(class_name_this))
class_names_tmp1 = [label_idx["label_names"][index] for index in class_indices1]
class_names_tmp2 = [class_names[index] for index in class_indices2]
assert class_names_tmp1 == class_names_tmp2
return class_indices2
base_classes = get_class_indices(label_idx["base_classes"])
base_classes_val = get_class_indices(label_idx["base_classes_1"])
base_classes_test = get_class_indices(label_idx["base_classes_2"])
novel_classes_val = get_class_indices(label_idx["novel_classes_1"])
novel_classes_test = get_class_indices(label_idx["novel_classes_2"])
return (
base_classes,
base_classes_val,
base_classes_test,
novel_classes_val,
novel_classes_test,
)
class ImageJitter:
def __init__(self, transformdict):
transformtypedict = dict(
Brightness=ImageEnhance.Brightness,
Contrast=ImageEnhance.Contrast,
Sharpness=ImageEnhance.Sharpness,
Color=ImageEnhance.Color,
)
self.transforms = [(transformtypedict[k], transformdict[k]) for k in transformdict]
def __call__(self, img):
out = img
randtensor = torch.rand(len(self.transforms))
for i, (transformer, alpha) in enumerate(self.transforms):
r = alpha * (randtensor[i] * 2.0 - 1.0) + 1
out = transformer(out).enhance(r).convert("RGB")
return out
class ImageNetBase(data.Dataset):
def __init__(self, split="train", size256=False, transform=None):
dataset_name = "ImageNet256" if size256 else "ImageNet"
assert (split in ("train", "val")) or (split.find("train_subset") != -1)
self.split = split
self.name = f"{dataset_name}_Split_" + self.split
data_dir = _IMAGENET256_DATASET_DIR if size256 else _IMAGENET_DATASET_DIR
print(f"==> Loading {dataset_name} dataset - split {self.split}")
print(f"==> {dataset_name} directory: {data_dir}")
self.transform = transform
print(f"==> transform: {self.transform}")
train_dir = os.path.join(data_dir, "train")
val_dir = os.path.join(data_dir, "val")
split_dir = train_dir if (self.split.find("train") != -1) else val_dir
self.data = datasets.ImageFolder(split_dir, self.transform)
self.labels = [item[1] for item in self.data.imgs]
if self.split.find("train_subset") != -1:
subsetK = int(self.split[len("train_subset") :])
assert subsetK > 0
self.split = "train"
label2ind = utils.build_label_index(self.data.targets)
all_indices = []
for label, img_indices in label2ind.items():
assert len(img_indices) >= subsetK
all_indices += img_indices[:subsetK]
self.data.imgs = [self.data.imgs[idx] for idx in all_indices]
self.data.samples = [self.data.samples[idx] for idx in all_indices]
self.data.targets = [self.data.targets[idx] for idx in all_indices]
self.labels = [self.labels[idx] for idx in all_indices]
def __getitem__(self, index):
img, label = self.data[index]
return img, label
def __len__(self):
return len(self.data)
class ImageNet(ImageNetBase):
def __init__(
self,
split="train",
use_geometric_aug=True,
use_simple_geometric_aug=False,
use_color_aug=True,
cutout_length=0,
do_not_use_random_transf=False,
size256=False,
):
transform_train = []
assert not (use_simple_geometric_aug and use_geometric_aug)
if use_geometric_aug:
transform_train.append(transforms.RandomResizedCrop(224))
transform_train.append(transforms.RandomHorizontalFlip())
elif use_simple_geometric_aug:
transform_train.append(transforms.Resize(256))
transform_train.append(transforms.RandomCrop(224))
transform_train.append(transforms.RandomHorizontalFlip())
else:
transform_train.append(transforms.Resize(256))
transform_train.append(transforms.CenterCrop(224))
if use_color_aug:
jitter_params = {"Brightness": 0.4, "Contrast": 0.4, "Color": 0.4}
transform_train.append(ImageJitter(jitter_params))
transform_train.append(lambda x: np.asarray(x))
transform_train.append(transforms.ToTensor())
transform_train.append(transforms.Normalize(mean=_MEAN_PIXEL, std=_STD_PIXEL))
if cutout_length > 0:
print(f"==> cutout_length: {cutout_length}")
transform_train.append(utils.Cutout(n_holes=1, length=cutout_length))
transform_train = transforms.Compose(transform_train)
self.transform_train = transform_train
transform_test = transforms.Compose(
[
transforms.Resize(256),
transforms.CenterCrop(224),
lambda x: np.asarray(x),
transforms.ToTensor(),
transforms.Normalize(mean=_MEAN_PIXEL, std=_STD_PIXEL),
]
)
if do_not_use_random_transf or split == "val":
transform = transform_test
else:
transform = transform_train
super().__init__(split=split, size256=size256, transform=transform)
class ImageNetLowShot(ImageNet):
def __init__(self, phase="train", split="train", do_not_use_random_transf=False):
assert phase in ("train", "test", "val")
assert split in ("train", "val")
use_aug = (phase == "train") and (do_not_use_random_transf == False)
super().__init__(split=split, use_geometric_aug=use_aug, use_color_aug=use_aug)
self.phase = phase
self.split = split
self.name = "ImageNetLowShot_Phase_" + phase + "_Split_" + split
print(f"==> Loading ImageNet few-shot benchmark - phase {phase}")
# ***********************************************************************
(
base_classes,
_,
_,
novel_classes_val,
novel_classes_test,
) = load_ImageNet_fewshot_split(self.data.classes)
# ***********************************************************************
self.label2ind = utils.build_label_index(self.labels)
self.labelIds = sorted(self.label2ind.keys())
self.num_cats = len(self.labelIds)
assert self.num_cats == 1000
self.labelIds_base = base_classes
self.num_cats_base = len(self.labelIds_base)
if self.phase == "val" or self.phase == "test":
self.labelIds_novel = (
novel_classes_val if (self.phase == "val") else novel_classes_test
)
self.num_cats_novel = len(self.labelIds_novel)
intersection = set(self.labelIds_base) & set(self.labelIds_novel)
assert len(intersection) == 0
class ImageNetLowShotFeatures:
def __init__(self, data_dir, image_split="train", phase="train"):
# data_dir: path to the directory with the saved ImageNet features.
# image_split: the image split of the ImageNet that will be loaded.
# phase: whether the dataset will be used for training, validating, or
# testing the few-shot model model.
assert image_split in ("train", "val")
assert phase in ("train", "val", "test")
self.phase = phase
self.image_split = image_split
self.name = (
f"ImageNetLowShotFeatures_ImageSplit_{self.image_split}" f"_Phase_{self.phase}"
)
dataset_file = os.path.join(data_dir, "ImageNet_" + self.image_split + ".h5")
self.data_file = h5py.File(dataset_file, "r")
self.count = self.data_file["count"][0]
self.features = self.data_file["all_features"][...]
self.labels = self.data_file["all_labels"][: self.count].tolist()
# ***********************************************************************
data_tmp = datasets.ImageFolder(os.path.join(_IMAGENET_DATASET_DIR, "train"), None)
(
base_classes,
base_classes_val,
base_classes_test,
novel_classes_val,
novel_classes_test,
) = load_ImageNet_fewshot_split(data_tmp.classes)
# ***********************************************************************
self.label2ind = utils.build_label_index(self.labels)
self.labelIds = sorted(self.label2ind.keys())
self.num_cats = len(self.labelIds)
assert self.num_cats == 1000
self.labelIds_base = base_classes
self.num_cats_base = len(self.labelIds_base)
if self.phase == "val" or self.phase == "test":
self.labelIds_novel = (
novel_classes_val if (self.phase == "val") else novel_classes_test
)
self.num_cats_novel = len(self.labelIds_novel)
intersection = set(self.labelIds_base) & set(self.labelIds_novel)
assert len(intersection) == 0
self.base_classes_eval_split = (
base_classes_val if (self.phase == "val") else base_classes_test
)
def __getitem__(self, index):
features_this = torch.Tensor(self.features[index]).view(-1, 1, 1)
label_this = self.labels[index]
return features_this, label_this
def __len__(self):
return int(self.count)