-
Notifications
You must be signed in to change notification settings - Fork 37
/
Experiment_HMDB51.py
194 lines (154 loc) · 6.53 KB
/
Experiment_HMDB51.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
__author__ = "Yinchong Yang"
__copyright__ = "Siemens AG, 2017"
__licencse__ = "MIT"
__version__ = "0.1"
"""
MIT License
Copyright (c) 2017 Siemens AG
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
"""
import os
import numpy as np
import pickle
import datetime
from keras.layers import Input, SimpleRNN, LSTM, GRU, Dense, Dropout, Masking, BatchNormalization
from keras.models import Model
from keras.optimizers import *
from keras.regularizers import l2
from keras.layers.wrappers import TimeDistributed
from keras.preprocessing.sequence import pad_sequences
from keras.utils.np_utils import to_categorical
from sklearn.metrics import average_precision_score, roc_auc_score, accuracy_score
# Custom Functions -----------------------------------------------------------------------------------------------------
from TTRNN import TT_GRU, TT_LSTM
data_path = ''
write_out_path = ''
splits_path = ''
def get_clips(class_name):
files = os.listdir(data_path + class_name)
files.sort()
clip_list = []
for this_file in files:
clip_list.append( data_path + class_name + '/' + this_file )
return clip_list
np.random.seed(11111986)
CV_setting = 1 # [1, 2, 3]
model_type = 0
use_TT = 1
classes = os.listdir(data_path)
classes.sort()
clips = [None]*51
labels = [None]*51
sizes = np.zeros(51)
X_train_lists = [None]*51
splits = [None]*51
for k in range(51):
this_clip = get_clips(classes[k])
clips[k] = this_clip
sizes[k] = len(this_clip)
labels[k] = np.repeat([k], sizes[k])
split_read = splits_path + classes[k] + '_split.txt'
splits[k] = np.genfromtxt(split_read, dtype=None)[:, CV_setting]
# flatten both lists
clips = np.array( [item for sublist in clips for item in sublist] )
splits = np.array( [item for sublist in splits for item in sublist] )
labels = np.array( [item for sublist in labels for item in sublist] )
labels = to_categorical(labels)
GLOBAL_MAX_LEN = 90
np.random.seed(11111986)
shuffle_inds = np.random.choice(range(len(clips)), len(clips), False)
clips = clips[shuffle_inds]
labels = labels[shuffle_inds]
CV_splits = np.array_split(np.arange(len(clips)), 5)
test_inds = CV_splits[CV_setting]
train_inds = np.setdiff1d(np.arange(len(clips)), test_inds)
train_set = [clips[train_inds], labels[train_inds]]
test_set = [clips[test_inds], labels[test_inds]]
def load_data(inds, mode = 'train', maxlen = GLOBAL_MAX_LEN):
N = len(inds)
X = np.zeros((N, maxlen, 120*160*3), dtype='int8')
if mode == 'train':
set = train_set
else:
set = test_set
for i in range(N):
read_in = open(set[0][inds[i]])
this_clip = pickle.load(read_in)[0] # of shape (nb_frames, 240, 320, 3)
read_in.close()
# take every 3rd frame due to memory restriction
this_clip = this_clip[np.arange(0, this_clip.shape[0], 3)] # of shape (nb_frames/3, 240, 320, 3)
# flatten the dimensions 1, 2 and 3
this_clip = this_clip.reshape(this_clip.shape[0], -1) # of shape (nb_frames/3, 240*320*3)
this_clip = (this_clip - 128).astype('int8') # this_clip.mean()
X[i] = pad_sequences([this_clip], maxlen=maxlen, truncating='post', dtype='int8')[0]
Y = set[1][inds]
return [X, Y]
n_tr = len(train_set[0])
n_te = len(test_set[0])
# test
# X_train, Y_train = load_data(np.arange(500),mode='train')
# X_test, Y_test = load_data(np.arange(100),mode='test')
alpha = 1e-2
tt_input_shape = [4, 20, 20, 36]
tt_output_shape = [4, 4, 4, 4]
tt_ranks = [1, 4, 4, 4, 1]
input = Input(shape=(GLOBAL_MAX_LEN, 120*160*3))
if model_type == 0:
if use_TT == 0:
rnn_layer = GRU(output_dim=225, activation='tanh',
dropout=0.25, recurrent_dropout=0.25, )
elif use_TT == 1:
rnn_layer = TT_GRU(tt_input_shape=tt_input_shape, tt_output_shape=tt_output_shape,
tt_ranks=tt_ranks,
dropout=0.25, recurrent_dropout=0.25, activation='tanh',
return_sequences=False, ortho_init=False )
elif model_type == 1:
if use_TT == 0:
rnn_layer = LSTM(output_dim=225, activation='tanh', consume_less='gpu',
dropout=0.25, recurrent_dropout=0.25, )
elif use_TT == 1:
rnn_layer = TT_LSTM(tt_input_shape=tt_input_shape, tt_output_shape=tt_output_shape,
tt_ranks=tt_ranks,
dropout=0.25, recurrent_dropout=0.25, activation='tanh',
return_sequences=False, ortho_init=False)
h = rnn_layer(input)
h = Dropout(0.25)(h)
output = Dense(output_dim=51, activation='softmax', kernel_regularizer=l2(alpha))(h)
model = Model(input, output)
model.compile(optimizer=Adam(1e-4), loss='categorical_crossentropy', metrics=['accuracy'])
if use_TT == 1:
compress_factor = rnn_layer.compress_factor
else:
compress_factor = 1
# X_train, Y_train = load_data(np.arange(0, 16), mode='train')
X_train, Y_train = load_data(np.arange(0, n_tr), mode='train')
# X_test, Y_test = load_data(np.arange(0, 16), mode='test')
X_test, Y_test = load_data(np.arange(0, n_te), mode='test')
file_name = str(CV_setting) + '_' + str(model_type) + '_' + str(use_TT)
start = datetime.datetime.now()
for l in range(401):
print 'iter ' + str(l)
model.fit(X_train, Y_train, epochs=1, batch_size=16, verbose=1, )
if l % 50 == 0:
res = model.evaluate(X_test, Y_test)
print 'Test res: '
print res
write_out = open(write_out_path + file_name + '_' + str(l) + '.pkl', 'wb')
pickle.dump([model.get_weights(), res, compress_factor], write_out)
write_out.close()
stop = datetime.datetime.now()
print stop-start