-
Notifications
You must be signed in to change notification settings - Fork 0
/
data_creation_patch.py
234 lines (163 loc) · 6.3 KB
/
data_creation_patch.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
# coding: utf-8
# In[30]:
get_ipython().magic(u'matplotlib inline')
import glob
import os
import numpy as np
import scipy.io as sio
import time
import pickle
import h5py
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import cv2
import random
from PIL import Image
# In[15]:
TYNAMO_HOME_DIR = '/nfs/tynamo/home/data/vision7/gdhody/chrono/'
BLITZLE_HOME_DIR = '/nfs/blitzle/home/data/vision5/gdhody/chrono/'
HOME_DIR = TYNAMO_HOME_DIR
HDF5_PATH = os.path.join(HOME_DIR, 'storage.hdf5')
PICKLE_PATH = os.path.join(HOME_DIR, 'storage.pkl')
# In[26]:
LABEL_FILE = os.path.join(HOME_DIR, 'UCF_HMDB_ACT.mat')
PICKLE_PATH = os.path.join(HOME_DIR, 'patch.pkl')
HDF5_PATH = os.path.join(HOME_DIR, 'patch.hdf5')
# In[17]:
TRAIN_DATA = sio.loadmat(LABEL_FILE)
# In[18]:
REMOVE_CLASSES = [
'MILITARYPARADE',
'HAIRCUT',
'TAICHI',
'YOYO',
'APPLYEYEMAKEUP',
'BABYCRAWLING',
'PLAYINGTABLA',
'WRITINGONBOARD',
'ROCKCLIMBINGINDOOR',
'BANDMARCHING',
'DRUMMING',
'PLAYINGDAF',
'BLOWINGCANDLES',
'PLAYINGDHOL',
'MOPPINGFLOOR',
'PLAYINGPIANO',
'TYPING',
'SKIJET',
'HEADMASSAGE',
'PLAYINGSITAR',
'HORSERACE',
'SKYDIVING',
'PLAYINGFLUTE',
'APPLYLIPSTICK',
'BRUSHINGTEETH',
'SURFING',
'JUGGLINGBALLS',
'PLAYINGGUITAR',
'SHAVINGBEARD',
'BILLIARDS',
'KNITTING',
'FENCING',
'BOXINGSPEEDBAG',
'MIXING',
'BLOWDRYER',
'SALSASPIN',
'CUTTINGINKITCHEN',
'RAFTING',
'HORSERACE',
]
# In[19]:
TRAIN_UCF = []
TRAIN_HMDB = []
for index, sample in enumerate(TRAIN_DATA['filename']):
image_file_name = sample[0][0]
if 'UCF101' in image_file_name:
image_class = image_file_name.split('/')[-1].split('_')[1].upper()
if image_class not in REMOVE_CLASSES:
image_path = os.path.join(HOME_DIR, image_file_name)
if len(glob.glob(image_path)):
frames, crop = TRAIN_DATA['frame'][index], TRAIN_DATA['crop'][index]
TRAIN_UCF.append((image_file_name, np.array(list(crop)), np.array(list(frames))))
if 'HMDB51' in image_file_name:
image_path = os.path.join(HOME_DIR, image_file_name)
if len(glob.glob(image_path)):
frames, crop = TRAIN_DATA['frame'][index], TRAIN_DATA['crop'][index]
TRAIN_HMDB.append((image_file_name, np.array(list(crop)), np.array(list(frames))))
# In[20]:
print len(TRAIN_UCF), len(TRAIN_HMDB)
random.shuffle(TRAIN_UCF)
random.shuffle(TRAIN_HMDB)
random.shuffle(TRAIN_UCF)
random.shuffle(TRAIN_HMDB)
# In[21]:
with open(PICKLE_PATH, 'wb') as f:
pickle.dump([TRAIN_UCF, TRAIN_HMDB], f, pickle.HIGHEST_PROTOCOL)
# In[22]:
with open(PICKLE_PATH, 'rb') as f:
TRAIN_UCF, TRAIN_HMDB = pickle.load(f)
# In[24]:
train_samples_ucf, validation_samples_ucf = (70 * len(TRAIN_UCF)) / 100, len(TRAIN_UCF) / 10
test_samples_ucf = len(TRAIN_UCF) - train_samples_ucf - validation_samples_ucf
train_samples_hmdb, validation_samples_hmdb = (70 * len(TRAIN_HMDB)) / 100, len(TRAIN_HMDB) / 10
test_samples_hmdb = len(TRAIN_HMDB) - train_samples_hmdb - validation_samples_hmdb
# In[35]:
train_shape_ucf = (train_samples_ucf, 4, 100, 100, 3)
val_shape_ucf = (validation_samples_ucf, 4, 100, 100, 3)
test_shape_ucf = (test_samples_ucf, 4, 100, 100, 3)
# In[36]:
hdf5_file = h5py.File(HDF5_PATH, mode='w')
hdf5_file.create_dataset("train_img", train_shape_ucf, np.float32)
hdf5_file.create_dataset("val_img", val_shape_ucf, np.float32)
hdf5_file.create_dataset("test_img", test_shape_ucf, np.float32)
# In[37]:
image_size, image_padding = 80, 20
for index, sample in enumerate(TRAIN_UCF[:train_samples_ucf]):
if index % 10000 == 0 and index > 1:
print 'Train data: {}/{}'.format(index, test_samples_ucf)
im_file_name, crop, frames = sample
im_file_paths = [os.path.join(HOME_DIR, im_file_name, 'Image' + str(frame) + '.jpg') for frame in frames]
images = [Image.open(im_file_path) for im_file_path in im_file_paths]
top_point = crop[0]
left_point = crop[1]
images = [image.crop((left_point, top_point,
left_point + image_size + image_padding,
top_point + image_size + image_padding))
for image in images]
images = [np.array(image, dtype=np.float32) for image in images]
img = np.stack((images[0], images[1], images[2], images[3]), axis=0)
hdf5_file["train_img"][index, ...] = img[None]
# In[38]:
image_size, image_padding = 80, 20
for index, sample in enumerate(TRAIN_UCF[train_samples_ucf:train_samples_ucf + validation_samples_ucf]):
if index % 10000 == 0 and index > 1:
print 'Train data: {}/{}'.format(index, validation_samples_ucf)
im_file_name, crop, frames = sample
im_file_paths = [os.path.join(HOME_DIR, im_file_name, 'Image' + str(frame) + '.jpg') for frame in frames]
images = [Image.open(im_file_path) for im_file_path in im_file_paths]
top_point = crop[0]
left_point = crop[1]
images = [image.crop((left_point, top_point,
left_point + image_size + image_padding,
top_point + image_size + image_padding))
for image in images]
images = [np.array(image, dtype=np.float32) for image in images]
img = np.stack((images[0], images[1], images[2], images[3]), axis=0)
hdf5_file["val_img"][index, ...] = img[None]
# In[40]:
image_size, image_padding = 80, 20
for index, sample in enumerate(TRAIN_UCF[train_samples_ucf + validation_samples_ucf:]):
if index % 5000 == 0 and index > 1:
print 'Train data: {}/{}'.format(index, test_samples_ucf)
im_file_name, crop, frames = sample
im_file_paths = [os.path.join(HOME_DIR, im_file_name, 'Image' + str(frame) + '.jpg') for frame in frames]
images = [Image.open(im_file_path) for im_file_path in im_file_paths]
top_point = crop[0]
left_point = crop[1]
images = [image.crop((left_point, top_point,
left_point + image_size + image_padding,
top_point + image_size + image_padding))
for image in images]
images = [np.array(image, dtype=np.float32) for image in images]
img = np.stack((images[0], images[1], images[2], images[3]), axis=0)
hdf5_file["test_img"][index, ...] = img[None]