-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathprocess_data.py
166 lines (140 loc) · 6.62 KB
/
process_data.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
import numpy as np
import pandas as pd
import AbstractModel as abst
import keras
from keras.models import Sequential, load_model
from keras.layers import Activation, BatchNormalization, Dense, Dropout, Flatten
from keras.layers import Conv2D, MaxPooling2D
from keras.optimizers import SGD
from sklearn.model_selection import train_test_split
class SymbolModel(abst.AbsModel):
def load_train(self, train, train_labels):
numbers = train_labels[train_labels['label'].isin(list(range(10, 13)))]
self.train = train[train['index'].isin(numbers['index'].values)].drop(['index'], axis=1)
self.train = self.train.as_matrix().reshape(numbers['index'].count(), 24, 24, 1)
self.train_labels = numbers.drop(['index'], axis=1).as_matrix()
one_hot = np.zeros([self.train_labels.shape[0], 3])
for i, label in enumerate(self.train_labels):
one_hot[i][label - 10] = 1
self.train_labels = one_hot
def load_model(self):
self.model = Sequential()
self.model.add(Conv2D(32, (3, 3), activation='relu', input_shape=(24, 24, 1)))
self.model.add(Conv2D(32, (3, 3), activation='relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2)))
self.model.add(Dropout(0.25))
self.model.add(Conv2D(64, (3, 3), activation='relu'))
self.model.add(Conv2D(64, (3, 3), activation='relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2)))
self.model.add(Dropout(0.25))
self.model.add(Flatten())
self.model.add(Dense(256, activation='relu'))
self.model.add(Dropout(0.5))
self.model.add(Dense(3, activation='softmax'))
self.model.compile(loss='mean_squared_error', optimizer='adam')
def load_model_from_file(self, filename):
self.model = load_model(filename)
def fit(self):
self.model.fit(self.train, self.train_labels, batch_size=32, epochs=10)
self.model.save('symbol_model.h5')
def score(self):
x_train, x_test, y_train, y_test = train_test_split(self.train, self.train_labels)
self.model.fit(x_train, y_train, batch_size=32, epochs=10)
score = self.model.evaluate(x_test, y_test, batch_size=32)
print(score)
def load_test(self, test):
test_no_index = test.drop(['index'], axis=1)
test_matrices = test_no_index.as_matrix().reshape(20000, 24, 120, 1)
test_ims = np.array([test_matrices[:, :, 24 * i:24 * (i + 1), :] for i in [1, 3]])
self.test = test_ims
def predict(self):
symbarr = ['+', '-', '=']
symbol1 = np.array([symbarr[symb] for symb in np.argmax(self.model.predict(self.test[0, :, :, :]), axis=1)])
symbol2 = np.array([symbarr[symb] for symb in np.argmax(self.model.predict(self.test[1, :, :, :]), axis=1)])
newdf = pd.DataFrame(np.array([symbol1, symbol2]).T, columns=["symbol1", "symbol2"])
newdf.to_csv("symbol.csv")
class NumberModel(abst.AbsModel):
def load_train(self, train, train_labels):
numbers = train_labels[train_labels['label'].isin(list(range(10)))]
self.train = train[train['index'].isin(numbers['index'].values)].drop(['index'], axis=1)
self.train = self.train.as_matrix().reshape(numbers['index'].count(), 24, 24, 1)
self.train_labels = numbers.drop(['index'], axis=1).as_matrix()
one_hot = np.zeros([self.train_labels.shape[0], 10])
for i, label in enumerate(self.train_labels):
one_hot[i][label] = 1
self.train_labels = one_hot
def load_model(self):
self.model = Sequential()
self.model.add(Conv2D(32, (3, 3), padding='same', input_shape=(24, 24, 1)))
self.model.add(BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(Dropout(0.25))
self.model.add(Conv2D(32, (3, 3), padding='same'))
self.model.add(BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2)))
self.model.add(Dropout(0.25))
self.model.add(Conv2D(64, (3, 3), padding='same'))
self.model.add(BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2)))
self.model.add(Dropout(0.25))
self.model.add(Conv2D(128, (3, 3), padding='same'))
self.model.add(BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2)))
self.model.add(Dropout(0.25))
self.model.add(Flatten())
self.model.add(Dense(500, use_bias = False))
self.model.add(BatchNormalization())
self.model.add(Activation('relu'))
self.model.add(Dropout(0.25))
self.model.add(Dense(10, activation='softmax'))
self.model.compile(loss='categorical_crossentropy', optimizer='adam')
def load_model_from_file(self, filename):
self.model = load_model(filename)
def fit(self):
self.log(str(self.model.to_json()))
self.model.fit(self.train, self.train_labels, batch_size=128, epochs=20)
self.model.save('number_model.h5')
def score(self):
x_train, x_test, y_train, y_test = train_test_split(self.train, self.train_labels)
self.model.fit(x_train, y_train, batch_size=32, epochs=10)
score = self.model.evaluate(x_test, y_test, batch_size=32)
print(score)
def load_test(self, test):
test_no_index = test.drop(['index'], axis=1)
test_matrices = test_no_index.as_matrix().reshape(20000, 24, 120, 1)
test_ims = np.array([test_matrices[:, :, 24 * i:24 * (i + 1), :] for i in range(0, 5, 2)])
self.test = test_ims
def predict(self):
num1scores = self.model.predict(self.test[0, :, :, :])
num2scores = self.model.predict(self.test[1, :, :, :])
num3scores = self.model.predict(self.test[2, :, :, :])
pd.DataFrame(num1scores).to_csv("num1scores.csv")
pd.DataFrame(num2scores).to_csv("num2scores.csv")
pd.DataFrame(num3scores).to_csv("num3scores.csv")
symbol_model = SymbolModel()
symbol_model.load_train(
pd.read_csv("data/train.csv"),
pd.read_csv("data/train_labels.csv")
)
symbol_model.load_model()
symbol_model.fit()
symbol_model.load_test(
pd.read_csv("data/test.csv")
)
symbol_model.load_model_from_file('symbol_model.h5')
symbol_model.predict()
number_model = NumberModel()
number_model.load_train(
pd.read_csv("data/train.csv"),
pd.read_csv("data/train_labels.csv")
)
number_model.load_model()
number_model.fit()
number_model.load_test(
pd.read_csv("data/test.csv")
)
number_model.load_model_from_file('number_model.h5')
number_model.predict()