example7_1_CNN.py

from keras.models import Model
from keras.layers import Input, Dense, Conv2D, MaxPool2D, Flatten
from keras.utils import to_categorical
import matplotlib.pyplot as plt
import numpy as np
import cv2

#Create model
input = Input(shape = (50,50,1))
conv1 = Conv2D(8,3,activation='relu')(input)
pool1 = MaxPool2D(pool_size=(2, 2))(conv1)
conv2 = Conv2D(8,3,activation='relu')(pool1)
pool2 = MaxPool2D(pool_size=(2, 2))(conv2)
flat = Flatten()(pool2)
hidden = Dense(16, activation='relu')(flat)
output = Dense(3, activation='softmax')(hidden)
model = Model(inputs=input, outputs=output)

model.compile(optimizer='adam',
              loss='categorical_crossentropy',
              metrics=['accuracy'])

model.summary()


#Read data from file
N = 30
x_train = np.zeros((N,50,50,1),'float')
y_train = np.zeros((N),'float')
count = 0
for class_id in range(1,4):
    for im_id in range(1,11):
        im = cv2.imread("thainum123/train/"+str(class_id)+"/"+str(im_id)+".bmp",cv2.IMREAD_GRAYSCALE)
        im = cv2.resize(im,(50,50))
        x_train[count,:,:,0] = im/255.
        y_train[count] = class_id-1
        count += 1

y_train = to_categorical(y_train)

#Train Model
h = model.fit(x_train, y_train, epochs=20)

plt.plot(h.history['acc'])
plt.show()


#Test Model
N = 15
x_test = np.zeros((N,50,50,1),'float')
y_test = np.zeros((N),'float')
count = 0
for class_id in range(1,4):
    for im_id in range(1,6):
        im = cv2.imread("thainum123/test/"+str(class_id)+"/"+str(im_id)+".bmp",cv2.IMREAD_GRAYSCALE)
        im = cv2.resize(im,(50,50))
        x_test[count,:,:,0] = im/255.
        y_test[count] = class_id-1
        count += 1

y_test = to_categorical(y_test)

score = model.evaluate(x_test, y_test)
print(score)

y_pred = model.predict(x_test)
print(y_pred)
print(np.argmax(y_pred,axis = -1)+1)