-
Notifications
You must be signed in to change notification settings - Fork 32
/
run2.py
184 lines (173 loc) · 5.1 KB
/
run2.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
import numpy as np
import os
from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Flatten, Activation
from keras.layers import Input
from keras.layers.convolutional import Convolution2D, MaxPooling2D, ZeroPadding2D, Convolution3D, ZeroPadding3D
from keras.layers.pooling import MaxPooling3D
from keras.optimizers import SGD, Adam
from keras.models import Model
from keras.utils import np_utils
from sklearn.metrics import mean_squared_error
import glob
from keras import __version__ as keras_version
np.random.seed(420)
def read_data ():
data = []
y1 = []
y2 = []
car = 0
cyc = 0
ped = 0
files = glob.glob ('grid_files/*.csv')
for i in range (len(files)):
files[i] = files [i].replace ('.csv', '')
f = files[i]
dt = np.loadtxt (f+'.csv', delimiter=',')
x = np.zeros (10*10*10*6)
x = x.reshape ((6,10,10,10))
for j in dt:
for k in range (6):
x[k][int(j[0]-1)][int(j[01]-1)][int(j[02]-1)] = j[k+3]
yd = x
dt = np.loadtxt (f+'.txt', delimiter=' ', dtype=np.str)
r = np.zeros (3)
for j in dt:
x = np.zeros (3)
if j[0] == 'Car' and r[0] != 1:
x[0] = 1
r[0] = 1
car += 1
y1.append (x)
x = j[11:]
x = np.array(x, dtype='|S24')
x = x.astype(np.float32)
y2.append (x)
data.append (yd)
elif j[0] == 'Cyclist' and r[01] != 1:
x[1] = 1
r[1] = 1
cyc += 1
y1.append (x)
x = j[11:]
x = np.array(x, dtype='|S24')
x = x.astype(np.float32)
y2.append (x)
data.append (yd)
elif j[0] == 'Pedestrian' and r[02] != 1:
x[2] = 1
r[2] = 1
ped += 1
y1.append (x)
x = j[11:]
x = np.array(x, dtype='|S24')
x = x.astype(np.float32)
y2.append (x)
data.append (yd)
data, y1, y2 = np.array (data), np.array (y1), np.array (y2)
arr = np.arange(len (data))
np.random.shuffle(arr)
TX = data[arr[0:len (arr)*0.6]]
tX = data[arr[len (arr)*0.6:len (arr)*0.895]]
vX = data[arr[len (arr)*0.895:]]
Ty1 = y1[arr[0:len (arr)*0.6]]
ty1 = y1[arr[len (arr)*0.6:len (arr)*0.895]]
vy1 = y1[arr[len (arr)*0.895:]]
Ty2 = y2[arr[0:len (arr)*0.6]]
ty2 = y2[arr[len (arr)*0.6:len (arr)*0.895]]
vy2 = y2[arr[len (arr)*0.895:]]
print 'Cars = ', car, 'Peds = ', ped, 'Cyc = ', cyc
return TX, Ty1, Ty2, tX, ty1, ty2, vX, vy1, vy2
def calc_err (a, b):
s = np.zeros (len(a))
for i in range (len (a)):
s[i] = ((a[i] - b[i]) ** 2)
def gen_model (input_shape=(6,10,10,10)):
model = Sequential()
inp = Input (input_shape)
model.add(Convolution3D(8, 3, 3, 3, border_mode='same',
input_shape=input_shape))
X = Convolution3D(8, 3, 3, 3, border_mode='same')(inp)
model.add(Activation('relu'))
X = Activation('relu') (X)
model.add(ZeroPadding3D())
X = ZeroPadding3D()(X)
model.add(Convolution3D(8, 3, 3, 3))
X = Convolution3D(8, 3, 3, 3)(X)
model.add(Activation('relu'))
X = Activation('relu')(X)
model.add(ZeroPadding3D())
X = ZeroPadding3D()(X)
model.add(Convolution3D(1, 3, 3, 3))
X = Convolution3D(1, 3, 3, 3)(X)
model.add(Activation('relu'))
X = Activation('relu')(X)
model.add(Flatten())
X = Flatten()(inp)#(X)
cm = model
cm.compile(optimizer=Adam (1e-4), loss='mse'
#, metrics = ['accuracy']
)
Z = X
model.add(Dense(1024))
X = Dense(1024)(X)
model.add(Activation('relu'))
X = Activation('relu')(X)
model.add(Dense(1024))
X = Dense(1024)(X)
model.add(Activation('relu'))
X = Activation('relu')(X)
'''model.add(Dense(3))
X1 = Dense(3)(X)
model.add(Activation('softmax'))
X1 = Activation('softmax', name='Class Act')(X1)
'''
model.add(Dense(400))
X2 = Dense(400)(X)
model.add(Activation('relu'))
X2 = Activation('relu')(X2)
model.add(Dense(4))
X2 = Dense(4)(X2)
model.add(Activation('relu'))
model.compile(optimizer=Adam (1e-4), loss='mse'
#, metrics = ['accuracy']
)
return model,cm
X2 = Activation('relu', name='Box Act')(X2)
model = Model (inp, [X1, X2])
convModel = Model (inp, Z)
sgd = SGD(lr=2e-3, decay=1e-4, momentum=0.9, nesterov=True)
adm = Adam (lr = 1e-4)
model.compile(optimizer=sgd, loss=['binary_crossentropy', 'mse']
#, metrics = ['accuracy']
)
convModel.compile(optimizer=sgd, loss='mse'
#, metrics = ['accuracy']
)
return model, convModel
if __name__ == '__main__':
print('Keras version: {}'.format(keras_version))
model, cm = gen_model ()
model.summary ()
print 'Reading data'
TX, Ty1, Ty2, tX, ty1, ty2, vX, vy1, vy2 = read_data ()
print (TX.shape)
print (Ty1.shape)
print (Ty2.shape)
print (tX.shape)
print (ty1.shape)
print (ty2.shape)
print (vX.shape)
print (vy1.shape)
print (vy2.shape)
#model.fit(TX, [Ty1, Ty2], batch_size=16, nb_epoch=11, validation_data=(vX, [vy1, vy2]))
model.fit(TX, Ty2, batch_size=16, nb_epoch=11, validation_data=(vX, vy2))
model.save_weights('vote3deep.h5')
#y1, y2 = model.predict(tX, batch_size=32, verbose=0)
y2 = model.predict(tX, batch_size=32, verbose=0)
y = cm.predict (TX[0:3])
np.savetxt ('0.txt', y)
#er1 = mean_squared_error (ty1, y1)
er2 = mean_squared_error (ty2, y2)
#print 'Total error 1 = ', er1
print 'Total error 2 = ', er2