-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathMaster.py
277 lines (235 loc) · 10.6 KB
/
Master.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
from src.NN_MODELS.cnn_lstm import *
from src.NN_MODELS.inception_v3 import *
from src.NN_MODELS.vgg_testing import *
from src.NN_MODELS.gann_ import *
#from src.NN_MODELS. import *
from src.common import *
#from src.NN_MODELS.network_tests import *
#from src.DATA_PREPARATION.partition_grouped import *
#from src.DATA_PREPARATION.partition_grouped_folders import *
import keras
from bayes_opt import BayesianOptimization
import time
#Location of Frame Differencing
FD_DATA_LOC = os.path.join("DATA","ambient_data","fd_data")#"1104_raw_full_ambient")#,"1104_FD_Full_Set")
#Location of Normal Data
NORMAL_DATA_LOC = os.path.join("DATA","ambient_data","raw_data")
##############################################################################
################# FRAME DIFF VS NON-FRAME DIFF ##############################
################# PREPROP VS NON-PREPROP ###################################
###########################################################################
IM_HEIGHT=250
FD = "True"
Preprop = "True"
conv1_size = 6
cached_model= None
moment = 0.9
decay = 6
dense_size = 8
conv2_size = 7
conv1_size = 6
lr=4
NUMBER_EPOCHS = 150
IM_WIDTH=100
file = open('MODEL_OUTPUTS/Dataset_Time_to_train.txt',mode='w')
file.write("Start,End,Difference\n")
start = time.time()
for FD_ in [0,1]:
if FD_ == 0:
FD = "False"
TRAIN_DATA = os.path.join(NORMAL_DATA_LOC,"training_data")
VALIDATE_DATA = os.path.join(NORMAL_DATA_LOC,"validation_data")
else:
FD = "True"
TRAIN_DATA = os.path.join(FD_DATA_LOC,"training_data")
VALIDATE_DATA = os.path.join(FD_DATA_LOC,"validation_data")
for prep_ in [0,1]:
if prep_ == 0:
Preprop = "False"
datagen = ImageDataGenerator(rescale=1. / 255)
datagenval = ImageDataGenerator(rescale= 1./255)
else:
Preprop = "True"
datagen = ImageDataGenerator(horizontal_flip=True,vertical_flip=True,width_shift_range=0.1,
height_shift_range=0.1,channel_shift_range=0.1,shear_range=0.1,
rotation_range = 10,rescale=1. / 255)
datagenval = ImageDataGenerator(rescale= 1./255)
for model in ["GANN","Vgg","Inception","CNN"]:
if model == "Inception":
vals = "LR-" + str(lr) + "_D-" + str(decay) + "_M-" + str(moment) + "_IM-" + str(IM_WIDTH) + "_FD-" + FD
vals += "_Time_Stamp-" + str(time.time()) + "_Preprop-" + Preprop
inc = INCEPTION_V3(lr=0.0001,cached_model= None,IM_HEIGHT=IM_HEIGHT,IM_WIDTH=IM_WIDTH)
inc.train(
train_directory_=TRAIN_DATA, validation_directory_= VALIDATE_DATA,
model_description= vals,epochs=NUMBER_EPOCHS,
datagen=datagen,datagenval=datagenval)
elif model == "Vgg":
vals = "LR-" + str(lr) + "_C1-" + str(conv1_size) + "_C2-" + str(conv2_size) + "_DS-" + str(dense_size)
vals += "_D-" + str(decay) + "_M-" + str(moment) + "_IM-" + str(IM_WIDTH) + "_FD-" + FD + "_Time_Stamp-" + str(time.time()) + "_Preprop-" + Preprop
vgg_ = VGG(IM_HEIGHT=IM_HEIGHT,IM_WIDTH=IM_WIDTH,lr=lr,conv1_size=conv1_size,conv2_size=conv2_size,dense_size=dense_size,decay=decay,moment=moment)
vgg_.train(train_directory_=TRAIN_DATA, validation_directory_=VALIDATE_DATA, model_description=vals,epochs=NUMBER_EPOCHS,datagen=datagen,datagenval=datagenval)
elif model == "CNN":
if FD_ == 0 and prep_ == 0:
vals = "LR-" + str(lr) + "_D-" + str(decay) + "_M-" + str(moment) + "_IM-" + str(IM_WIDTH) + "_FD-" + FD
vals += "_Time_Stamp-" + str(time.time()) + "_Preprop-" + Preprop
cnn = CNN_LSTM(lr=0.01)
cnn.train(train_directory_= TRAIN_DATA, validation_directory_=VALIDATE_DATA, model_description=vals, epochs=NUMBER_EPOCHS)
elif model == "GANN":
if prep_ == 0:
GANN = WGANN()
GANN.train(train_directory_= TRAIN_DATA, validation_directory_=VALIDATE_DATA, model_description = "Gann",epochs=NUMBER_EPOCHS)
end = time.time()
diff = end-start
file.write(str(start) + ',' + str(end) + ',' + str(diff) + '\n')
file.close()
##############################################################################
################# TEST IMAGE SIZE ###########################################
############################################################################
'''
datagen = ImageDataGenerator(
horizontal_flip=True,
vertical_flip=True,
width_shift_range=0.1,
height_shift_range=0.1,
channel_shift_range=0.1,
shear_range=0.1,
rotation_range = 10,
rescale=1. / 255)
datagenval = ImageDataGenerator(rescale= 1./255)
NUMBER_EPOCHS = 75
for FD_ in [1]:
if FD_ == 0:
FD = "False"
TRAIN_DATA = os.path.join(NORMAL_DATA_LOC,"training_data")
VALIDATE_DATA = os.path.join(NORMAL_DATA_LOC,"validation_data")
else:
FD = "True"
TRAIN_DATA = os.path.join(FD_DATA_LOC,"training_data")
VALIDATE_DATA = os.path.join(FD_DATA_LOC,"validation_data")
file = open('MODEL_OUTPUTS/image_sizetime.txt',mode='w')
file.write("Image_Size,Start,End,Difference\n")
for image_size in [50,100,150,200,250,300,350,400]:
IM_HEIGHT=image_size
Preprop = "True"
conv1_size = 6
cached_model= None
moment = 0.9
decay = 6
dense_size = 8
conv2_size = 7
conv1_size = 6
lr=4
start = time.time()
#STRING TO SAVE MODEL AS
vals = "LR-" + str(lr) + "_C1-" + str(conv1_size) + "_C2-" + str(conv2_size) + "_DS-" + str(dense_size)
vals += "_D-" + str(decay) + "_M-" + str(moment) + "_IM-" + str(image_size) + "_FD-" + FD + "_Time_Stamp-" + str(time.time()) + "_Preprop-" + Preprop
vgg_ = VGG(False,IM_HEIGHT=IM_HEIGHT,IM_WIDTH=IM_WIDTH,lr=lr,conv1_size=conv1_size,conv2_size=conv2_size,dense_size=dense_size,decay=decay,moment=moment)
vgg_.train(train_directory_=TRAIN_DATA, validation_directory_=VALIDATE_DATA, model_description= vals,
epochs=NUMBER_EPOCHS,datagen=datagen,datagenval=datagenval)
end = time.time()
diff = end-start
file.write(str(image_size) + ',' + str(start) + ',' + str(end) + ',' + str(diff) + '\n')
file.close()
'''
##############################################################################
################# OPTIMISE NETWORK ###########################################
############################################################################
'''
IM_HEIGHT=250
FD = "True"
Preprop = "True"
conv1_size = 6
cached_model= None
moment = 0.9
decay = 6
dense_size = 8
conv2_size = 7
conv1_size = 6
lr=4
NUMBER_EPOCHS = 50
IM_WIDTH=250
datagen = ImageDataGenerator(
horizontal_flip=True,
vertical_flip=True,
width_shift_range=0.1,
height_shift_range=0.1,
channel_shift_range=0.1,
shear_range=0.1,
rotation_range = 10,
rescale=1. / 255)
datagenval = ImageDataGenerator(rescale= 1./255)
def get_vals(conv1_size = 5,conv2_size = 6,dense_size = 8,moment = 0.9):
global lr
global decay
print("Dense Size is " + str(dense_size))
vals = "LR-" + str(lr) + "_C1-" + str(conv1_size) + "_C2-" + str(conv2_size) + "_DS-" + str(dense_size)
vals += "_D-" + str(decay) + "_M-" + str(moment) + "_IM-" + str(IM_WIDTH) + "_FD-" + FD + "_Time_Stamp-" + str(time.time()) + "_Preprop-" + Preprop
print(vals)
TRAIN_DATA = os.path.join(FD_DATA_LOC,"training_data")
VALIDATE_DATA = os.path.join(FD_DATA_LOC,"validation_data")
try:
vgg_ = VGG(False,IM_HEIGHT=IM_HEIGHT,IM_WIDTH=IM_WIDTH,lr=lr,conv1_size=conv1_size,conv2_size=conv2_size,dense_size=dense_size,decay=decay,moment=moment)
vgg_.train(train_directory_=TRAIN_DATA, validation_directory_=VALIDATE_DATA,
model_description="optimise" + str(vals), epochs=NUMBER_EPOCHS, datagen=datagen,
datagenval=datagenval)
validate_generator = datagenval.flow_from_directory(
VALIDATE_DATA,
target_size=(IM_HEIGHT, IM_WIDTH),
batch_size=BATCH_SIZE,
class_mode="categorical")
print(vgg_.model.metrics_names)
data = vgg_.model.evaluate_generator(validate_generator)
print(data)
return data[1]
except:
return 0.0
#RETURNS A GRID BETWEEN TWO VALUES
def return_grid(value,dist):
grid = np.arange(value[0],value[1]+dist/10,dist/10,dtype=np.float64)
return grid
'''
##############################################################################
################# LR vs LR_DECAY GRID #######################################
############################################################################
'''
#Run a 2D grid of the network to determine best parameters
def minimise_lr_and_lr_decay():
global kwargs
global lr
global decay
global NUMBER_EPOCHS
#NUMBER_EPOCHS = 30
max_lr = None
max_lr_dec = None
max_val = None
dist = 6-3
grid = return_grid([3,6], dist)
dist2 = 8-0.1
grid2 = return_grid([0.1,8], dist)
print("LR_GRID is " + str(grid))
print("LR_Decay is " + str(grid2))
for lr_decay in grid2:
for lr_ in grid:
values = []
max_item = None
max_val = 0
decay = lr_decay
lr = lr_
values.append(get_vals())
if values[-1]>max_val:
max_val = values[-1]
max_lr = lr_
max_lr_dec = lr_decay
print ("Maximum Val was : " + str(max_val) + " Lr " + str(max_lr) + " lr_dec " + str(max_lr_dec))
NUMBER_EPOCHS = 70
minimise_lr_and_lr_decay()
'''
##############################################################################
################# BAYESIAN OPTIMISATION #####################################
############################################################################
'''
bo = BayesianOptimization(lambda conv1_size,conv2_size,dense_size,moment: get_vals(conv1_size,conv2_size,dense_size,moment),
{"conv1_size":(3,8),"conv2_size":(3,8),"dense_size":(6,11),"moment":(0.5,1.0)})
bo.explore({"conv1_size":(3,8),"conv2_size":(3,8),"dense_size":(6,11),"moment":(0.5,1.0)})
bo.maximize(init_points=2, n_iter=500, kappa=10,acq="ucb") #, acq="ucb"
'''