def train(self):
msg = QtWidgets.QMessageBox()
load_train = LoadData()
if self.lblHasil.text() != "":
self.lblHasil.setText("")
if self.txtLR.text() == "" or self.txtEpoch.text() == "":
# Show Message
msg.setIcon(msg.Information)
msg.setWindowTitle("Peringatan")
msg.setText("Isi Learning Rate atau Epoch terlebih dahulu!")
msg.exec_()
else:
# print(self.txtEpoch.text())
# print(self.txtLR.text())
self.data_train = load_train.data_train
self.train_generator, self.x_train, self.x_valid, self.y_train, self.y_valid = load_train.loadDataTrain(
)
self.input_shape = (load_train.img_rows, load_train.img_cols, 1)
epo = self.txtEpoch.text()
epo = int(epo)
lr = self.txtLR.text()
lr = float(lr)
batch_size = 32
cnn = CNN(self.input_shape,
len(self.train_generator.class_indices), lr)
model = cnn.ConvNetModel()
self.progressBar.setMinimum(0)
self.progressBar.setMaximum(epo)
history = model.fit(self.x_train,
self.y_train,
batch_size=batch_size,
epochs=epo,
verbose=1,
validation_data=(self.x_valid, self.y_valid))
self.progressBar.setValue(epo)
evaluation = model.evaluate(self.x_valid,
self.y_valid,
batch_size=batch_size,
verbose=1)
print('loss : %.2f, accuracy : %.2f' %
(evaluation[0], evaluation[1]))
self.lblHasil.setText('Hasil Pelatihan: '
'Loss : %.2f, '
'Accuracy : %.2f%%' %
(evaluation[0], evaluation[1] * 100))
model.save('./model.h5')
plt.plot(history.history['acc'])
plt.plot(history.history['val_acc'])
plt.title('Akurasi Model')
plt.ylabel('Akurasi')
plt.xlabel('Epoch')
plt.legend(['train', 'test'], loc='upper right')
plt.show()
