-
Notifications
You must be signed in to change notification settings - Fork 8
/
Copy pathdataset.py
177 lines (150 loc) · 7.26 KB
/
dataset.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
# import mmcv
import decord
from PIL import Image, ImageEnhance
import torch
import numpy as np
import torchvision.transforms as transforms
import json
import os
identity = lambda x:x
transformtypedict=dict(Brightness=ImageEnhance.Brightness, Contrast=ImageEnhance.Contrast, Sharpness=ImageEnhance.Sharpness, Color=ImageEnhance.Color)
class SubVdieoDataset:
def __init__(self, sub_meta, cl, transform=transforms.ToTensor(), target_transform=identity, random_select=False, num_segments=None):
self.sub_meta = sub_meta
# self.video_list = [x.strip().split(' ') for x in open(sub_meta)]
if True:
self.image_tmpl = 'img_{:05d}.jpg'
else:
self.image_tmpl = 'img_{:05d}.png'
self.cl = cl
self.transform = transform
self.target_transform = target_transform
self.random_select = random_select
self.num_segments = num_segments
def __getitem__(self,i):
# image_path = os.path.join( self.sub_meta[i])
assert len(self.sub_meta[i]) == 2
full_path = self.sub_meta[i][0]
num_frames = self.sub_meta[i][1]
num_segments = self.num_segments
if self.random_select and num_frames>8 : # random sample
# frame_id = np.random.randint(num_frames)
average_duration = num_frames // num_segments
frame_id = np.multiply(list(range(num_segments)), average_duration)
frame_id = frame_id + np.random.randint(average_duration, size=num_segments)
else:
# frame_id = num_frames//2
tick = num_frames / float(num_segments)
frame_id = np.array([int(tick / 2.0 + tick * x) for x in range(num_segments)])
frame_id = frame_id + 1 # idx >= 1
img_group = []
for k in range(self.num_segments):
img_path = os.path.join(full_path,self.image_tmpl.format(frame_id[k]))
img = Image.open(img_path)
img = self.transform(img)
img_group.append(img)
img_group = torch.stack(img_group,0)
target = self.target_transform(self.cl)
# print('ok',image_path)
return img_group, target
def __len__(self):
return len(self.sub_meta)
class SetDataManager:
def __init__(self, image_size, n_way, n_support, n_query, num_segments, n_eposide =100):
super(SetDataManager, self).__init__()
self.image_size = image_size
self.n_way = n_way
self.batch_size = n_support + n_query
self.n_eposide = n_eposide
self.trans_loader = TransformLoader(image_size)
self.num_segments = num_segments
def get_data_loader(self, data_file, aug): #parameters that would change on train/val set
transform = self.trans_loader.get_composed_transform(aug)
dataset = SetDataset( data_file , self.batch_size, transform, random_select=aug, num_segments=self.num_segments) # video
sampler = EpisodicBatchSampler(len(dataset), self.n_way, self.n_eposide )
data_loader_params = dict(batch_sampler = sampler, num_workers = 8, pin_memory = True)
data_loader = torch.utils.data.DataLoader(dataset, **data_loader_params)
return data_loader
class TransformLoader:
def __init__(self, image_size,
# normalize_param = dict(mean= [0.485, 0.456, 0.406] , std=[0.229, 0.224, 0.225]),
normalize_param = dict(mean= [0.376, 0.401, 0.431] , std=[0.224, 0.229, 0.235]),
jitter_param = dict(Brightness=0.4, Contrast=0.4, Color=0.4)):
self.image_size = image_size
self.normalize_param = normalize_param
self.jitter_param = jitter_param
def parse_transform(self, transform_type):
if transform_type=='ImageJitter':
method = ImageJitter( self.jitter_param )
return method
method = getattr(transforms, transform_type)
if transform_type=='RandomResizedCrop':
return method(self.image_size)
elif transform_type=='CenterCrop':
return method(self.image_size)
elif transform_type=='Resize':
return method([int(self.image_size*1.15), int(self.image_size*1.15)])
elif transform_type=='Normalize':
return method(**self.normalize_param )
else:
return method()
def get_composed_transform(self, aug = False):
if aug:
transform_list = ['RandomResizedCrop', 'ImageJitter', 'RandomHorizontalFlip', 'ToTensor', 'Normalize']
else:
transform_list = ['Resize','CenterCrop', 'ToTensor', 'Normalize']
transform_funcs = [ self.parse_transform(x) for x in transform_list]
transform = transforms.Compose(transform_funcs)
return transform
class SetDataset: # frames
def __init__(self, data_file, batch_size, transform, random_select=False, num_segments=None):
# with open(data_file, 'r') as f:
# self.meta = json.load(f)
self.video_list = [x.strip().split(' ') for x in open(data_file)]
# self.cl_list = np.unique(self.meta['image_labels']).tolist()
self.cl_list = np.zeros(len(self.video_list),dtype=int)
for i in range(len(self.video_list)):
self.cl_list[i] = self.video_list[i][2]
self.cl_list = np.unique(self.cl_list).tolist()
self.sub_meta = {}
for cl in self.cl_list:
self.sub_meta[cl] = []
# for x,y in zip(self.meta['image_names'],self.meta['image_labels']):
# self.sub_meta[y].append(x)
for x in range(len(self.video_list)):
root_path = self.video_list[x][0]
num_frames = int(self.video_list[x][1])
label = int(self.video_list[x][2])
self.sub_meta[label].append([root_path,num_frames])
self.sub_dataloader = []
sub_data_loader_params = dict(batch_size = batch_size,
shuffle = True,
num_workers = 0, #use main thread only or may receive multiple batches
pin_memory = False)
for cl in self.cl_list:
sub_dataset = SubVdieoDataset(self.sub_meta[cl], cl, transform = transform ,random_select = random_select, num_segments=num_segments)
self.sub_dataloader.append( torch.utils.data.DataLoader(sub_dataset, **sub_data_loader_params) )
def __getitem__(self,i):
return next(iter(self.sub_dataloader[i]))
def __len__(self):
return len(self.cl_list)
class EpisodicBatchSampler(object):
def __init__(self, n_classes, n_way, n_episodes):
self.n_classes = n_classes
self.n_way = n_way
self.n_episodes = n_episodes
def __len__(self):
return self.n_episodes
def __iter__(self):
for i in range(self.n_episodes):
yield torch.randperm(self.n_classes)[:self.n_way]
class ImageJitter(object):
def __init__(self, transformdict):
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