def test_image_data_generator_training():
np.random.seed(1337)
img_gen = ImageDataGenerator(rescale=1.) # Dummy ImageDataGenerator
input_shape = (16, 16, 3)
(x_train, y_train), (x_test, y_test) = get_test_data(num_train=500,
num_test=200,
input_shape=input_shape,
classification=True,
num_classes=4)
y_train = to_categorical(y_train)
y_test = to_categorical(y_test)
model = Sequential([
layers.Conv2D(filters=8, kernel_size=3,
activation='relu',
input_shape=input_shape),
layers.MaxPooling2D(pool_size=2),
layers.Conv2D(filters=4, kernel_size=(3, 3),
activation='relu', padding='same'),
layers.GlobalAveragePooling2D(),
layers.Dense(y_test.shape[-1], activation='softmax')
])
model.compile(loss='categorical_crossentropy',
optimizer='rmsprop',
metrics=['accuracy'])
history = model.fit_generator(img_gen.flow(x_train, y_train, batch_size=16),
epochs=10,
validation_data=img_gen.flow(x_test, y_test,
batch_size=16),
verbose=0)
assert history.history['val_acc'][-1] > 0.75
model.evaluate_generator(img_gen.flow(x_train, y_train, batch_size=16))
def train(dataReader, oneHot, oneHotAveraged, contextHashes): n = (Epochs + 1) * SamplesPerEpoch # TODO + 1 should not be needed tokeniser = Tokenizer(nb_words=MaxWords) tokeniser.fit_on_texts((row[0] for row in dataReader.trainingData(n))) # `word_index` maps each word to its unique index dictionarySize = len(tokeniser.word_index) + 1 oneHotDimension = (1 if oneHotAveraged else SequenceLength) * dictionarySize if oneHot else 0 contextHashesDimension = dictionarySize * 2 if contextHashes else 0 model = Sequential() model.add(Dense(EmbeddingDim, input_dim=(oneHotDimension + contextHashesDimension))) model.add(Dense(Labels, activation='softmax')) model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy']) trainingGenerator = mapGenerator(dataReader.trainingData(n), tokeniser, dictionarySize, oneHot, oneHotAveraged, contextHashes) validationGenerator = mapGenerator(dataReader.validationData(n), tokeniser, dictionarySize, oneHot, oneHotAveraged, contextHashes) model.fit_generator(trainingGenerator, nb_epoch=Epochs, samples_per_epoch=SamplesPerEpoch, validation_data=validationGenerator, nb_val_samples=SamplesPerEpoch) model2 = Sequential() model2.add(Dense(EmbeddingDim, input_dim=(oneHotDimension + contextHashesDimension), weights=model.layers[0].get_weights())) return model, model2, tokeniser, dictionarySize
def train_CAE():
encoder = containers.Sequential()
encoder.add(Permute((3,1,2),input_shape=(h,w,ch))) # reorder input to ch, h, w (no sample axis)
encoder.add(GaussianNoise(0.05)) # corrupt inputs slightly
encoder.add(Convolution2D(16,3,3,init='glorot_uniform',border_mode='same'))
encoder.add(MaxPooling2D((2,2)))
encoder.add(Activation('tanh'))
encoder.add(Convolution2D(32,3,3,init='glorot_uniform',border_mode='same'))
encoder.add(MaxPooling2D((2,2)))
encoder.add(Activation('tanh'))
decoder = containers.Sequential()
decoder.add(UpSampling2D((2,2),input_shape=(32,32,32)))
decoder.add(Convolution2D(3,3,3,init='glorot_uniform',border_mode='same'))
decoder.add(Activation('tanh'))
decoder.add(UpSampling2D((2,2),input_shape=(16,64,64)))
decoder.add(Convolution2D(3,3,3,init='glorot_uniform',border_mode='same'))
decoder.add(Activation('tanh'))
decoder.add(Permute((2,3,1)))
autoencoder = AutoEncoder(encoder,decoder)
model = Sequential()
model.add(autoencoder)
model.compile(optimizer='rmsprop', loss='mae')
# if shapes don't match, check the output_shape of encoder/decoder
genr = image_generator(biz_id_train['photo_id'], batch_size)
model.fit_generator(genr, samples_per_epoch=len(biz_id_train), nb_epoch=10)
class CNN(object):
def __init__(self):
self.model = Sequential([
Conv2D(50, (3, 3), input_shape=(28, 28, 1), padding='same', activation='relu'),
Conv2D(50, (3, 3), input_shape=(28, 28, 1), padding='same', activation='relu'),
MaxPool2D(pool_size=(4, 4), strides=(3, 3), padding='same'),
Conv2D(32, (3, 3), padding='same', activation='relu' ),
Conv2D(32, (3, 3), padding='same', activation='relu' ),
MaxPool2D(pool_size=(7, 7), strides=(3, 3), padding='same'),
Flatten(),
Dropout(0.5),
Dense(64, activation='relu'),
Dense(10, activation='softmax')
])
self.model.compile(loss=categorical_crossentropy, optimizer=Adam(), metrics=['accuracy'])
def train(self):
mnist = input_data.read_data_sets("../MNIST_data/", one_hot=True)
train_datagen = ImageDataGenerator(rotation_range=20, width_shift_range=0.2, height_shift_range=0.2)
train_datagen.fit(mnist.train.images.reshape(-1, 28, 28, 1))
x_test, y_test = mnist.test.images.reshape(-1, 28, 28, 1), mnist.test.labels
self.model.fit_generator(train_datagen.flow(mnist.train.images.reshape(-1, 28, 28, 1), mnist.train.labels),
#batch_size=128,
epochs=20,
verbose=1,
validation_data=(x_test, y_test),
callbacks=[TrainValTensorBoard(log_dir='./logs/cnn4', histogram_freq=1, write_grads=True)])
score = self.model.evaluate(x_test, y_test, verbose=0)
print('Loss', score[0], 'acc', score[1])
def test_CallbackValData():
np.random.seed(1337)
(X_train, y_train), (X_test, y_test) = get_data_callbacks()
y_test = np_utils.to_categorical(y_test)
y_train = np_utils.to_categorical(y_train)
model = Sequential()
model.add(Dense(num_hidden, input_dim=input_dim, activation='relu'))
model.add(Dense(num_classes, activation='softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='sgd',
metrics=['accuracy'])
cbk = callbacks.LambdaCallback(on_train_end=lambda x: 1)
model.fit(X_train, y_train, batch_size=batch_size,
validation_data=(X_test, y_test), callbacks=[cbk], epochs=1)
cbk2 = callbacks.LambdaCallback(on_train_end=lambda x: 1)
train_generator = data_generator(X_train, y_train, batch_size)
model.fit_generator(train_generator, len(X_train), epochs=1,
validation_data=(X_test, y_test),
callbacks=[cbk2])
# callback validation data should always have x, y, and sample weights
assert len(cbk.validation_data) == len(cbk2.validation_data) == 3
assert cbk.validation_data[0] is cbk2.validation_data[0]
assert cbk.validation_data[1] is cbk2.validation_data[1]
assert cbk.validation_data[2].shape == cbk2.validation_data[2].shape
def test_multiprocessing_fit_error():
batch_size = 10
good_batches = 3
def custom_generator():
"""Raises an exception after a few good batches"""
for i in range(good_batches):
yield (np.random.randint(batch_size, 256, (50, 2)),
np.random.randint(batch_size, 2, 50))
raise RuntimeError
model = Sequential()
model.add(Dense(1, input_shape=(2, )))
model.compile(loss='mse', optimizer='adadelta')
samples = batch_size * (good_batches + 1)
with pytest.raises(StopIteration):
model.fit_generator(
custom_generator(), samples, 1,
workers=4, use_multiprocessing=True,
)
with pytest.raises(StopIteration):
model.fit_generator(
custom_generator(), samples, 1,
use_multiprocessing=False,
)
def main():
ext = extension_from_parameters()
out_dim = 1
loss = 'mse'
metrics = None
#metrics = ['accuracy'] if CATEGORICAL else None
datagen = RegressionDataGenerator()
train_gen = datagen.flow(batch_size=BATCH_SIZE)
val_gen = datagen.flow(val=True, batch_size=BATCH_SIZE)
val_gen2 = datagen.flow(val=True, batch_size=BATCH_SIZE)
model = Sequential()
for layer in LAYERS:
if layer:
model.add(Dense(layer, input_dim=datagen.input_dim, activation=ACTIVATION))
if DROP:
model.add(Dropout(DROP))
model.add(Dense(out_dim))
model.summary()
model.compile(loss=loss, optimizer='rmsprop', metrics=metrics)
train_samples = int(datagen.n_train/BATCH_SIZE) * BATCH_SIZE
val_samples = int(datagen.n_val/BATCH_SIZE) * BATCH_SIZE
history = BestLossHistory(val_gen2, val_samples, ext)
checkpointer = ModelCheckpoint(filepath='model'+ext+'.h5', save_best_only=True)
model.fit_generator(train_gen, train_samples,
nb_epoch = NB_EPOCH,
validation_data = val_gen,
nb_val_samples = val_samples,
callbacks=[history, checkpointer])
def test_multiprocessing_fit_error():
batch_size = 32
good_batches = 5
def myGenerator():
"""Raises an exception after a few good batches"""
for i in range(good_batches):
yield (np.random.randint(batch_size, 256, (500, 2)),
np.random.randint(batch_size, 2, 500))
raise RuntimeError
model = Sequential()
model.add(Dense(1, input_shape=(2, )))
model.compile(loss='mse', optimizer='adadelta')
samples = batch_size * (good_batches + 1)
with pytest.raises(Exception):
model.fit_generator(
myGenerator(), samples, 1,
nb_worker=4, pickle_safe=True,
)
with pytest.raises(Exception):
model.fit_generator(
myGenerator(), samples, 1,
pickle_safe=False,
)
class MLP(BaseEstimator):
def __init__(self, verbose=0, model=None, final_activation='sigmoid'):
self.verbose = verbose
self.model = model
self.final_activation = final_activation
def fit(self, X, y):
if not self.model:
self.model = Sequential()
self.model.add(Dense(1000, input_dim=X.shape[1]))
self.model.add(Activation('relu'))
self.model.add(Dropout(0.5))
self.model.add(Dense(y.shape[1]))
self.model.add(Activation(self.final_activation))
self.model.compile(loss='categorical_crossentropy', optimizer=Adam(lr=0.01))
self.model.fit_generator(generator=_batch_generator(X, y, 256, True),
samples_per_epoch=X.shape[0], nb_epoch=20, verbose=self.verbose)
def predict(self, X):
pred = self.predict_proba(X)
return sparse.csr_matrix(pred > 0.2)
def predict_proba(self, X):
pred = self.model.predict_generator(generator=_batch_generatorp(X, 512), val_samples=X.shape[0])
return pred
def CNN(trainDir, validationDir, classNum):
model = Sequential()
model.add(Convolution2D(4, 3, 3, input_shape=(img_width, img_height, 1)))
model.add(Activation('relu'))
model.add(Convolution2D(4, 3, 3))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
# layer
model.add(Convolution2D(8, 3, 3))
model.add(Activation('relu'))
model.add(Convolution2D(8, 3, 3))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
# model.add(Convolution2D(16, 3, 3))
# model.add(Activation('relu'))
# model.add(MaxPooling2D(pool_size=(2, 2)))
# layer
model.add(Flatten())
model.add(Dense(64))
model.add(Activation('relu'))
# model.add(Dropout(0.5))
model.add(Dense(16))
model.add(Activation('relu'))
model.add(Dropout(0.6))
model.add(Dense(classNum))
model.add(Activation('softmax'))
# test
model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
# this is the augmentation configuration we will use for training
train_datagen = ImageDataGenerator(
rescale=1./255,
shear_range=0.2,
zca_whitening=True,
zoom_range=0.2,
horizontal_flip=False)
# this is the augmentation configuration we will use for testing:
# only rescaling
test_datagen = ImageDataGenerator(rescale=1./255, zca_whitening=True)
train_generator = train_datagen.flow_from_directory(
trainDir,
target_size=(img_width, img_height),
batch_size=32,
color_mode='grayscale',
class_mode='categorical')
validation_generator = test_datagen.flow_from_directory(
validationDir,
target_size=(img_width, img_height),
batch_size=32,
color_mode='grayscale',
class_mode='categorical')
model.fit_generator(
train_generator,
samples_per_epoch=nb_train_samples,
nb_epoch=nb_epoch,
validation_data=validation_generator,
nb_val_samples=nb_validation_samples)
return model
def try_params( n_iterations, params, data=None, datamode='memory'):
print "iterations:", n_iterations
print_params( params )
batchsize = 100
if datamode == 'memory':
X_train, Y_train = data['train']
X_valid, Y_valid = data['valid']
inputshape = X_train.shape[1:]
else:
train_generator = data['train']['gen_func'](batchsize, data['train']['path'])
valid_generator = data['valid']['gen_func'](batchsize, data['valid']['path'])
train_epoch_step = data['train']['n_sample'] / batchsize
valid_epoch_step = data['valid']['n_sample'] / batchsize
inputshape = data['train']['gen_func'](batchsize, data['train']['path']).next()[0].shape[1:]
model = Sequential()
model.add(Conv2D(128, (1, 24), padding='same', input_shape=inputshape, activation='relu'))
model.add(GlobalMaxPooling2D())
model.add(Dense(32,activation='relu'))
model.add(Dropout(params['DROPOUT']))
model.add(Dense(2))
model.add(Activation('softmax'))
optim = Adadelta
myoptimizer = optim(epsilon=params['DELTA'], rho=params['MOMENT'])
mylossfunc = 'categorical_crossentropy'
model.compile(loss=mylossfunc, optimizer=myoptimizer,metrics=['accuracy'])
early_stopping = EarlyStopping( monitor = 'val_loss', patience = 3, verbose = 0 )
if datamode == 'memory':
model.fit(
X_train,
Y_train,
batch_size=batchsize,
epochs=int( round( n_iterations )),
validation_data=(X_valid, Y_valid),
callbacks = [ early_stopping ])
score, acc = model.evaluate(X_valid,Y_valid)
else:
model.fit_generator(
train_generator,
steps_per_epoch=train_epoch_step,
epochs=int( round( n_iterations )),
validation_data=valid_generator,
validation_steps=valid_epoch_step,
callbacks = [ early_stopping ])
score, acc = model.evaluate_generator(valid_generator, steps=valid_epoch_step)
return { 'loss': score, 'model': (model.to_json(), optim, myoptimizer.get_config(), mylossfunc) }
def lm():
maxlen=10
cfig = getattr(config, 'get_config_morph')('cs')
batch_size, nb_epoch = cfig['batch_size'], 200
X_train, y_train = getTextFile(cfig['train_file'], cfig['train_dic'], cfig)
X_train = sequence.pad_sequences(X_train, maxlen=maxlen, padding='post')
y_train = sequence.pad_sequences(y_train, maxlen=maxlen, padding='post')
#X_train, y_train = X_train[:10050], y_train[:10050]
print X_train.shape, y_train.shape
#X_test, y_test = getTextFile(cfig['test_file'], cfig['train_dic'], cfig)
#X_test = sequence.pad_sequences(X_test, maxlen=10, padding='post')
#y_test = sequence.pad_sequences(y_test, maxlen=10, padding='post')
'''
y_train_tensor3 = np.zeros((len(X_train), maxlen, cfig['vocab_size']), dtype=np.bool)
i, t = 0, 0
for sentence in y_train:
t = 0
for v in sentence:
y_train_tensor3[i][t][v] = True
t += 1
i += 1
k = 0
for i , j in generate_data(X_train, y_train, 200, cfig['vocab_size']):
print i.shape , j.shape
if k > 20:
break
k += 1
exit(0)
'''
print 'Build model...'
model = Sequential()
model.add(Embedding(cfig['vocab_size'], 128, dropout=0.2))
model.add(LSTM(128, return_sequences=True)) #- original
model.add(TimeDistributedDense(cfig['vocab_size']))
model.add(Activation('softmax'))
model.compile(loss='categorical_crossentropy', optimizer='rmsprop')
model.summary()
print 'Train...'
model.fit_generator(generate_data(X_train, y_train, batch_size, cfig['vocab_size']),
#samples_per_epoch=len(X_train)/batch_size,
samples_per_epoch=1000,
nb_epoch=nb_epoch)
#model.fit(X_train, y_train_tensor3)
exit(0)
cnt = 0
for i , j in generate_data(X_train, y_train, 200, cfig['vocab_size']):
#model.train_on_batch(i, j)
history= model.fit(i, j, batch_size=10, nb_epoch=1,verbose=0)
if cnt >= 3:
break
cnt += 1
def train_model(genre, dir_model, MP):
sess = tf.Session(config=tf.ConfigProto(log_device_placement=True)) #check gpu is being used
batch_size = MP['bs']
lstm_size = MP['lstm_size']
seq_length = MP['seq_length']
drop = MP['dropout']
lr = MP['lr']
epochs = MP['epochs']
text_to_int, int_to_text, n_chars = np.load('playlists/%s/ancillary_char.npy'%genre)
vocab_size = len(text_to_int)
X = np.load('playlists/%s/X_sl%d_char.npy'%(genre, seq_length))
y = np.load('playlists/%s/y_sl%d_char.npy'%(genre, seq_length))
# randomly shuffle samples before test/valid split
np.random.seed(40)
ran = [i for i in range(len(X))]
np.random.shuffle(ran)
X_train, X_valid, y_train, y_valid = train_test_split(X[ran], y[ran], test_size=0.2, random_state=42)
try:
model = load_model(dir_model)
print("successfully loaded previous model, continuing to train")
except:
print("generating new model")
model = Sequential()
model.add(GRU(lstm_size, dropout=drop, recurrent_dropout=drop, return_sequences=True,
input_shape=(seq_length, vocab_size)))
for i in range(MP['n_layers'] - 1):
model.add(GRU(lstm_size, dropout=drop, recurrent_dropout=drop, return_sequences=True))
model.add(TimeDistributed(Dense(vocab_size, activation='softmax'))) #output shape=(bs, sl, vocab)
decay = 0.5*lr/epochs
optimizer = Adam(lr=lr, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=decay, clipvalue=1)
#optimizer = RMSprop(lr=lr, decay=decay)
model.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=['categorical_accuracy'])
print(model.summary())
# callbacks
checkpoint = ModelCheckpoint(dir_model, monitor='loss', save_best_only=True, mode='min')
#earlystop = EarlyStopping(monitor='val_loss', min_delta=0.01, patience=3)
callbacks_list = [checkpoint]
# train
model.fit_generator(one_hot_gen(X_train, y_train, vocab_size, seq_length, batch_size),
steps_per_epoch=len(X_train)/batch_size, epochs=epochs, callbacks=callbacks_list,
validation_data=one_hot_gen(X_valid, y_valid, vocab_size, seq_length, batch_size),
validation_steps=len(X_valid)/batch_size)
model.save(dir_model)
def test_sequential_fit_generator():
(X_train, y_train), (X_test, y_test) = _get_test_data()
def data_generator(train):
if train:
max_batch_index = len(X_train) // batch_size
else:
max_batch_index = len(X_test) // batch_size
i = 0
while 1:
if train:
yield (X_train[i * batch_size: (i + 1) * batch_size], y_train[i * batch_size: (i + 1) * batch_size])
else:
yield (X_test[i * batch_size: (i + 1) * batch_size], y_test[i * batch_size: (i + 1) * batch_size])
i += 1
i = i % max_batch_index
model = Sequential()
model.add(Dense(nb_hidden, input_shape=(input_dim,)))
model.add(Activation('relu'))
model.add(Dense(nb_class))
model.add(Activation('softmax'))
model.compile(loss='categorical_crossentropy', optimizer='rmsprop')
model.fit_generator(data_generator(True), len(X_train), nb_epoch, show_accuracy=False)
model.fit_generator(data_generator(True), len(X_train), nb_epoch, show_accuracy=True)
model.fit_generator(data_generator(True), len(X_train), nb_epoch, show_accuracy=False, validation_data=(X_test, y_test))
model.fit_generator(data_generator(True), len(X_train), nb_epoch, show_accuracy=True, validation_data=(X_test, y_test))
loss = model.evaluate(X_train, y_train, verbose=0)
assert(loss < 0.9)
def test_sequential_fit_generator():
(x_train, y_train), (x_test, y_test) = _get_test_data()
def data_generator(train):
if train:
max_batch_index = len(x_train) // batch_size
else:
max_batch_index = len(x_test) // batch_size
i = 0
while 1:
if train:
yield (x_train[i * batch_size: (i + 1) * batch_size], y_train[i * batch_size: (i + 1) * batch_size])
else:
yield (x_test[i * batch_size: (i + 1) * batch_size], y_test[i * batch_size: (i + 1) * batch_size])
i += 1
i = i % max_batch_index
model = Sequential()
model.add(Dense(num_hidden, input_shape=(input_dim,)))
model.add(Activation('relu'))
model.add(Dense(num_class))
model.pop()
model.add(Dense(num_class))
model.add(Activation('softmax'))
model.compile(loss='categorical_crossentropy', optimizer='rmsprop')
model.fit_generator(data_generator(True), 5, epochs)
model.fit_generator(data_generator(True), 5, epochs,
validation_data=(x_test, y_test))
model.fit_generator(data_generator(True), 5, epochs,
validation_data=data_generator(False),
validation_steps=3)
model.fit_generator(data_generator(True), 5, epochs, max_queue_size=2)
model.evaluate(x_train, y_train)
def model(datagen, X_train, Y_train, X_test, Y_test):
batch_size = 32
nb_epoch = 200
# input image dimensions
img_rows, img_cols = 32, 32
# the CIFAR10 images are RGB
img_channels = 3
model = Sequential()
model.add(Convolution2D(32, 3, 3, border_mode='same',
input_shape=(img_channels, img_rows, img_cols)))
model.add(Activation('relu'))
model.add(Convolution2D(32, 3, 3))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout({{uniform(0, 1)}}))
model.add(Convolution2D(64, 3, 3, border_mode='same'))
model.add(Activation('relu'))
model.add(Convolution2D(64, 3, 3))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout({{uniform(0, 1)}}))
model.add(Flatten())
model.add(Dense(512))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(nb_classes))
model.add(Activation('softmax'))
# let's train the model using SGD + momentum (how original).
sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
model.compile(loss='categorical_crossentropy',
optimizer=sgd,
metrics=['accuracy'])
# fit the model on the batches generated by datagen.flow()
model.fit_generator(datagen.flow(X_train, Y_train,
batch_size=batch_size),
samples_per_epoch=X_train.shape[0],
nb_epoch=nb_epoch,
validation_data=(X_test, Y_test))
score, acc = model.evaluate(X_test, Y_test, verbose=0)
return {'loss': -acc, 'status': STATUS_OK, 'model': model}
def test_multiprocessing_training_fromfile():
reached_end = False
arr_data = np.random.randint(0,256, (500, 200))
arr_labels = np.random.randint(0, 2, 500)
np.savez("data.npz", **{"data": arr_data, "labels": arr_labels})
def myGenerator():
batch_size = 32
n_samples = 500
arr = np.load("data.npz")
while True:
batch_index = np.random.randint(0, n_samples - batch_size)
start = batch_index
end = start + batch_size
X = arr["data"][start: end]
y = arr["labels"][start: end]
yield X, y
# Build a NN
model = Sequential()
model.add(Dense(10, input_shape=(200, )))
model.add(Activation('relu'))
model.add(Dense(1))
model.add(Activation('linear'))
model.compile(loss='mse', optimizer='adadelta')
model.fit_generator(myGenerator(),
samples_per_epoch=320,
nb_epoch=1,
verbose=1,
max_q_size=10,
nb_worker=2,
pickle_safe=True)
model.fit_generator(myGenerator(),
samples_per_epoch=320,
nb_epoch=1,
verbose=1,
max_q_size=10,
pickle_safe=False)
reached_end = True
assert reached_end
def test_TerminateOnNaN():
np.random.seed(1337)
(X_train, y_train), (X_test, y_test) = get_data_callbacks()
y_test = np_utils.to_categorical(y_test)
y_train = np_utils.to_categorical(y_train)
cbks = [callbacks.TerminateOnNaN()]
model = Sequential()
initializer = initializers.Constant(value=1e5)
for _ in range(5):
model.add(Dense(num_hidden, input_dim=input_dim, activation='relu',
kernel_initializer=initializer))
model.add(Dense(num_classes, activation='linear'))
model.compile(loss='mean_squared_error',
optimizer='rmsprop')
# case 1 fit
history = model.fit(X_train, y_train, batch_size=batch_size,
validation_data=(X_test, y_test), callbacks=cbks, epochs=20)
loss = history.history['loss']
assert len(loss) == 1
assert loss[0] == np.inf
history = model.fit_generator(data_generator(X_train, y_train, batch_size),
len(X_train),
validation_data=(X_test, y_test),
callbacks=cbks,
epochs=20)
loss = history.history['loss']
assert len(loss) == 1
assert loss[0] == np.inf or np.isnan(loss[0])
def Convolution(test, df, flip_indices):
img_rows, img_cols = 96, 96 # images are 96x96 pixels
img_channels = 1 # images are grey scale - if RGB use img_channels = 3
nb_filter = 32 # common and efficient to use multiples of 2 for filters
nb_epoch = 1
batch_size = 32
X, y = load(df, test)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=99)
model = Sequential()
model.add(Convolution2D(nb_filter = nb_filter, nb_row = 3, nb_col = 3,
# border_mode = 'same', # this causes a crash, even though default is 'same'....
input_shape = X_train.shape[1:])) # must set input shape for first call to Convolutional2D
model.add(Activation('relu')) # default convolutional activation function - should try Leaky ReLU and PReLU functions : http://arxiv.org/abs/1502.01852
model.add(MaxPooling2D()) # pooling 2x2 with stride 2 is default - reduces size of matrix by half in both dimensions
# There is a new paper quesitoning the use of max pooling - finding that replacement with convolutional layers with increased stride works better
model.add(Dropout(0.1))
model.add(Convolution2D(nb_filter = nb_filter * 2, nb_row = 2, nb_col = 2))
model.add(Activation('relu'))
model.add(MaxPooling2D())
model.add(Dropout(0.2))
model.add(Convolution2D(nb_filter = nb_filter * 4, nb_row = 2, nb_col = 2))
model.add(Activation('relu'))
model.add(MaxPooling2D())
model.add(Dropout(0.3))
model.add(Flatten()) # flatten the data before fully connected layer (reg. neural net)
model.add(Dense(1024))
model.add(Activation('relu'))
model.add(Dropout(0.4))
model.add(Dense(1024))
model.add(Activation('relu'))
model.add(Dropout(0.5))
if len(flip_indices) == 12:
model.add(Dense(30))
else:
model.add(Dense(8))
model.compile(loss = 'mean_squared_error', # other objectives : http://keras.io/objectives/
optimizer = 'Adam') # discussion : http://cs231n.github.io/neural-networks-3/ : "Adam is currently recommended as the default algorithm to use" - cs231n
dataGen = createDataGen(flip_indices = flip_indices, horizontal_flip = True)
checkpointer = ModelCheckpoint(filepath="/tmp/weights.hdf5", verbose=1, save_best_only=True)
earlystopping = EarlyStopping(monitor = 'val_loss', patience = 20, verbose = 0, mode = 'auto')
model.fit_generator(dataGen.flow(X_train, y_train, batch_size = batch_size), nb_epoch = nb_epoch,
samples_per_epoch = X_train.shape[0],
validation_data = (X_test, y_test),
callbacks = [checkpointer, earlystopping])
class TrainAllCnn(object):
def __init__(self):
# prepare trainigs data
(X_train, y_train), (X_test, y_test) = cifar10.load_data()
print('X_train shape:', X_train.shape)
print(X_train.shape[0], 'train samples')
print(X_test.shape[0], 'test samples')
self.Y_train = np_utils.to_categorical(y_train, 10)
self.Y_test = np_utils.to_categorical(y_test, 10)
self.X_train = X_train.astype('float32')
self.X_test = X_test.astype('float32')
self.X_train /= 255
self.X_test /= 255
def build_model(self):
# we want an sequential model
self.model = Sequential()
self.model.add(Convolution2D(96, 3, 3, border_mode='same', input_shape=(3, 32, 32)))
self.model.add(Activation('relu'))
self.model.add(Convolution2D(96, 3, 3, border_mode='same'))
self.model.add(Activation('relu'))
self.model.add(Convolution2D(96, 3, 3, subsample=(2,2), border_mode='same'))
self.model.add(Activation('relu'))
self.model.add(Convolution2D(192, 3, 3, border_mode='same'))
self.model.add(Activation('relu'))
self.model.add(Convolution2D(192, 3, 3, border_mode='same'))
self.model.add(Activation('relu'))
self.model.add(Convolution2D(96, 3, 3, subsample=(2, 2), border_mode='same'))
self.model.add(Activation('relu'))
self.model.add(MaxPooling2D(pool_size=(3, 3), strides=(2,2)))
self.model.add(Convolution2D(192, 3, 3))
self.model.add(Activation('relu'))
self.model.add(Convolution2D(192, 1, 1))
self.model.add(Activation('relu'))
self.model.add(Convolution2D(10, 1, 1))
self.model.add(Activation('relu'))
self.model.add(Flatten())
self.model.add(Activation('softmax'))
# now we compile the model, asoptimizer we use stochastik gradient decent with momentum
sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
self.model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])
def train(self, batch, epoch):
self.build_model()
# now we fit the model
# with 200 epochs this takes a while ...
return self.model.fit(self.X_train, self.Y_train, batch_size=batch, nb_epoch=epoch, validation_data=(self.X_test, self.Y_test), shuffle=True)
def train_argument_images(self, batch, epoch):
self.build_model()
datagen = ImageDataGenerator(featurewise_center=False, samplewise_center=False, featurewise_std_normalization=False, samplewise_std_normalization=False, zca_whitening=False, rotation_range=0, width_shift_range=0.1, height_shift_range=0.1, horizontal_flip=True, vertical_flip=False, zoom_range=0.1)
datagen.fit(self.X_train)
return self.model.fit_generator(datagen.flow(self.X_train, self.Y_train, batch_size=batch), samples_per_epoch=self.X_train.shape[0], nb_epoch=epoch, validation_data=(self.X_test, self.Y_test)).model
def test_CallbackValData():
np.random.seed(1337)
(X_train, y_train), (X_test, y_test) = get_test_data(num_train=train_samples,
num_test=test_samples,
input_shape=(input_dim,),
classification=True,
num_classes=num_class)
y_test = np_utils.to_categorical(y_test)
y_train = np_utils.to_categorical(y_train)
model = Sequential()
model.add(Dense(num_hidden, input_dim=input_dim, activation='relu'))
model.add(Dense(num_class, activation='softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='sgd',
metrics=['accuracy'])
cbk = callbacks.LambdaCallback(on_train_end=lambda x: 1)
model.fit(X_train, y_train, batch_size=batch_size,
validation_data=(X_test, y_test), callbacks=[cbk], epochs=1)
def data_generator(train):
if train:
max_batch_index = len(X_train) // batch_size
else:
max_batch_index = len(X_test) // batch_size
i = 0
while 1:
if train:
yield (X_train[i * batch_size: (i + 1) * batch_size],
y_train[i * batch_size: (i + 1) * batch_size])
else:
yield (X_test[i * batch_size: (i + 1) * batch_size],
y_test[i * batch_size: (i + 1) * batch_size])
i += 1
i = i % max_batch_index
cbk2 = callbacks.LambdaCallback(on_train_end=lambda x: 1)
model.fit_generator(data_generator(True), len(X_train), epochs=1,
validation_data=(X_test, y_test),
callbacks=[cbk2])
# callback validation data should always have x, y, and sample weights
assert len(cbk.validation_data) == len(cbk2.validation_data) == 3
assert cbk.validation_data[0] is cbk2.validation_data[0]
assert cbk.validation_data[1] is cbk2.validation_data[1]
assert cbk.validation_data[2].shape == cbk2.validation_data[2].shape
def main():
ext = extension_from_parameters()
out_dim = 1
loss = 'mse'
metrics = None
#metrics = ['accuracy'] if CATEGORICAL else None
reshape = LOCALLY_CONNECTED_LAYERS is not None
datagen = RegressionDataGenerator()
train_gen = datagen.flow(batch_size=BATCH_SIZE, reshape=reshape)
val_gen = datagen.flow(val=True, batch_size=BATCH_SIZE, reshape=reshape)
val_gen2 = datagen.flow(val=True, batch_size=BATCH_SIZE, reshape=reshape)
model = Sequential()
if LOCALLY_CONNECTED_LAYERS:
for layer in LOCALLY_CONNECTED_LAYERS:
if layer:
model.add(LocallyConnected1D(*layer, input_shape=(datagen.input_dim, 1), activation=ACTIVATION))
if POOL:
model.add(MaxPooling1D(pool_length=POOL))
model.add(Flatten())
for layer in DENSE_LAYERS:
if layer:
model.add(Dense(layer, input_dim=datagen.input_dim, activation=ACTIVATION))
if DROP:
model.add(Dropout(DROP))
model.add(Dense(out_dim))
model.summary()
model.compile(loss=loss, optimizer='sgd', metrics=metrics)
train_samples = int(datagen.n_train/BATCH_SIZE) * BATCH_SIZE
val_samples = int(datagen.n_val/BATCH_SIZE) * BATCH_SIZE
history = BestLossHistory(val_gen2, val_samples, ext)
checkpointer = ModelCheckpoint(filepath='model'+ext+'.h5', save_best_only=True)
model.fit_generator(train_gen, train_samples,
nb_epoch = NB_EPOCH,
validation_data = val_gen,
nb_val_samples = val_samples,
callbacks=[history, checkpointer])
def test_multiprocessing_training():
reached_end = False
arr_data = np.random.randint(0, 256, (500, 2))
arr_labels = np.random.randint(0, 2, 500)
def myGenerator():
batch_size = 32
n_samples = 500
while True:
batch_index = np.random.randint(0, n_samples - batch_size)
start = batch_index
end = start + batch_size
X = arr_data[start: end]
y = arr_labels[start: end]
yield X, y
# Build a NN
model = Sequential()
model.add(Dense(1, input_shape=(2, )))
model.compile(loss='mse', optimizer='adadelta')
model.fit_generator(myGenerator(),
samples_per_epoch=320,
nb_epoch=1,
verbose=1,
max_q_size=10,
nb_worker=4,
pickle_safe=True)
model.fit_generator(myGenerator(),
samples_per_epoch=320,
nb_epoch=1,
verbose=1,
max_q_size=10,
pickle_safe=False)
reached_end = True
assert reached_end
def train(train_generator,train_size,input_num,dims_num):
print("Start Train Job! ")
start=time.time()
inputs=InputLayer(input_shape=(input_num,dims_num),batch_size=batch_size)
layer1=LSTM(128)
output=Dense(2,activation="softmax",name="Output")
optimizer=Adam()
model=Sequential()
model.add(inputs)
model.add(layer1)
model.add(Dropout(0.5))
model.add(output)
call=TensorBoard(log_dir=log_dir,write_grads=True,histogram_freq=1)
model.compile(optimizer,loss="categorical_crossentropy",metrics=["accuracy"])
model.fit_generator(train_generator,steps_per_epoch=train_size//batch_size,epochs=epochs_num,callbacks=[call])
# model.fit_generator(train_generator, steps_per_epoch=5, epochs=5, callbacks=[call])
model.save(model_dir)
end=time.time()
print("Over train job in %f s"%(end-start))
class SequenceTaggingMachine(object):
def __init__(self, mask ):
self.model = None
self.mask = mask
def train(self, train_corpus, valid_corpus, learning_param):
self.model = Sequential()
self.model.add(Embedding(train_corpus.source_cell_num(), 256, mask_zero = True))
self.model.add(Bidirectional(LSTM(128, return_sequences=True)))
self.model.add(Bidirectional(LSTM(128, return_sequences=True)))
self.model.add(TimeDistributed(Dense(input_dim=128, output_dim=train_corpus.target_cell_num())))
self.model.add(Activation('softmax'))
self.model.compile(loss=lambda output,target: masked_categorical_crossentropy(output, target, self.mask),
optimizer='rmsprop',
metrics=[lambda y_true, y_pred: masked_categorical_accuracy(y_true, y_pred, self.mask)])
logging.debug("Preparing data iter")
problem = SequenceTaggingProblem(train_corpus)
data_train = BucketIter(problem, learning_param.batch_size, max_pad_num=learning_param.max_pad)
val_problem = SequenceTaggingProblem(valid_corpus)
data_val = BucketIter(val_problem, learning_param.batch_size, max_pad_num=learning_param.max_pad)
checkpointer = ModelCheckpoint(filepath="weights.{epoch:03d}-{val_loss:.2f}.hdf5", verbose=1)
logging.debug("Begin train model")
self.model.fit_generator(bucket_iter_adapter(data_train,train_corpus.target_cell_num()),
samples_per_epoch=train_corpus.corpus_size(), nb_epoch=100, verbose=1,
validation_data=bucket_iter_adapter(data_val, train_corpus.target_cell_num()),
nb_val_samples = valid_corpus.corpus_size(), callbacks=[checkpointer])
print "Model is trained"
def test_multiprocessing_training_fromfile(in_tmpdir):
arr_data = np.random.randint(0, 256, (50, 2))
arr_labels = np.random.randint(0, 2, 50)
np.savez('data.npz', **{'data': arr_data, 'labels': arr_labels})
def custom_generator():
batch_size = 10
n_samples = 50
arr = np.load('data.npz')
while True:
batch_index = np.random.randint(0, n_samples - batch_size)
start = batch_index
end = start + batch_size
X = arr['data'][start: end]
y = arr['labels'][start: end]
yield X, y
# Build a NN
model = Sequential()
model.add(Dense(1, input_shape=(2, )))
model.compile(loss='mse', optimizer='adadelta')
model.fit_generator(custom_generator(),
steps_per_epoch=5,
epochs=1,
verbose=1,
max_queue_size=10,
workers=2,
use_multiprocessing=True)
model.fit_generator(custom_generator(),
steps_per_epoch=5,
epochs=1,
verbose=1,
max_queue_size=10,
use_multiprocessing=False)
os.remove('data.npz')
def rnn_gru(float_data, lookback=1440, step=6):
model = Sequential()
model.add(layers.GRU(32, input_shape=(None, float_data.shape[-1])))
model.add(layers.Dense(1))
model.compile(optimizer=RMSprop(), loss='mae')
history = model.fit_generator(train_gen,
steps_per_epoch=500,
epochs=20,
validation_data=val_gen,
validation_steps=val_steps)
draw_histroy(history)
def test_multithreading_from_file():
arr_data = np.random.randint(0, 256, (50, 2))
arr_labels = np.random.randint(0, 2, 50)
np.savez('data_threads.npz', **{'data': arr_data, 'labels': arr_labels})
@threadsafe_generator
def custom_generator():
batch_size = 10
n_samples = 50
arr = np.load('data_threads.npz')
while True:
batch_index = np.random.randint(0, n_samples - batch_size)
start = batch_index
end = start + batch_size
X = arr['data'][start: end]
y = arr['labels'][start: end]
yield X, y
# Build a NN
model = Sequential()
model.add(Dense(1, input_shape=(2,)))
model.compile(loss='mse', optimizer='adadelta')
# - Produce data on 4 worker threads, consume on main thread:
# - All worker threads share the SAME generator
model.fit_generator(custom_generator(),
steps_per_epoch=STEPS_PER_EPOCH,
epochs=1,
verbose=1,
validation_steps=None,
max_queue_size=10,
workers=WORKERS,
use_multiprocessing=False)
# - Produce data on 1 worker thread, consume on main thread:
# - Worker thread is the only thread running the generator
model.fit_generator(custom_generator(),
steps_per_epoch=STEPS_PER_EPOCH,
epochs=1,
verbose=1,
validation_steps=None,
max_queue_size=10,
workers=1,
use_multiprocessing=False)
# - Produce and consume data without a queue on main thread
# - Make sure the value of `use_multiprocessing` is ignored
model.fit_generator(custom_generator(),
steps_per_epoch=STEPS_PER_EPOCH,
epochs=1,
verbose=1,
validation_steps=None,
max_queue_size=10,
workers=0,
use_multiprocessing=False)
os.remove('data_threads.npz')
def train_xy(epochs=50, batch_size=32, h=256, w=256, ch=3, train_p=0.8, valid_p=0.1):
print("Compiling Model")
t_comp = time()
model = Sequential()
# reshape input to ch, h, w (no sample axis)
model.add(Reshape(dims=(h, w, ch), input_shape=(ch * h * w,)))
model.add(Permute((3, 1, 2)))
# add conv layers
model.add(Convolution2D(16, 3, 3, init="glorot_uniform", activation="relu", subsample=(1, 1)))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Convolution2D(32, 3, 3, init="glorot_uniform", activation="relu", subsample=(1, 1)))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Convolution2D(64, 3, 3, init="glorot_uniform", activation="relu", subsample=(1, 1)))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Convolution2D(64, 3, 3, init="glorot_uniform", activation="relu", subsample=(1, 1)))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Flatten())
model.add(Dense(output_dim=2000, init="glorot_uniform", activation="relu", W_regularizer=l2(0.001)))
model.add(Dropout(0.5))
model.add(Dense(output_dim=2000, init="glorot_uniform", activation="relu", W_regularizer=l2(0.001)))
model.add(Dropout(0.5))
model.add(Dense(output_dim=4, init="glorot_uniform", activation="relu"))
model.compile(optimizer="rmsprop", loss="mse")
t_train = time()
print("Took %.1fs" % (t_train - t_comp))
# split dataset
i_test = int(train_p * nrow) / batch_size * batch_size
i_valid = int(i_test * (1 - valid_p)) / batch_size * batch_size
X_train, Y_train = X_[:i_valid,], Y_[:i_valid,]
X_valid, Y_valid = X_[i_valid:i_test,], Y_[i_valid:i_test,]
# naive fitting to lower rmse faster
hist = model.fit(X_train, Y_train, batch_size=batch_size, nb_epoch=10, verbose=1, validation_split=0.1)
print(hist)
# fit by batch using generator!
img_aug = image_augmentor(X_, Y_, i_valid)
hist = model.fit_generator(
generator=img_aug,
samples_per_epoch=i_valid,
nb_epoch=5000,
verbose=1,
validation_data=(X_valid, Y_valid),
nb_worker=1,
)
rmse_test = model.evaluate(X_[i_test:,], Y_[i_test:,])
print("Test RMSE: %.4f" % rmse_test)
# save model
model_json = model.to_json()
open(path_img + "locate/model_116.json", "w").write(model_json)
model.save_weights(path_img + "locate/model_116_weights.h5")
def test_stop_training_csv(tmpdir):
np.random.seed(1337)
fp = str(tmpdir / 'test.csv')
(X_train, y_train), (X_test, y_test) = get_test_data(num_train=train_samples,
num_test=test_samples,
input_shape=(input_dim,),
classification=True,
num_classes=num_classes)
y_test = np_utils.to_categorical(y_test)
y_train = np_utils.to_categorical(y_train)
cbks = [callbacks.TerminateOnNaN(), callbacks.CSVLogger(fp)]
model = Sequential()
for _ in range(5):
model.add(Dense(num_hidden, input_dim=input_dim, activation='relu'))
model.add(Dense(num_classes, activation='linear'))
model.compile(loss='mean_squared_error',
optimizer='rmsprop')
def data_generator():
i = 0
max_batch_index = len(X_train) // batch_size
tot = 0
while 1:
if tot > 3 * len(X_train):
yield (np.ones([batch_size, input_dim]) * np.nan,
np.ones([batch_size, num_classes]) * np.nan)
else:
yield (X_train[i * batch_size: (i + 1) * batch_size],
y_train[i * batch_size: (i + 1) * batch_size])
i += 1
tot += 1
i = i % max_batch_index
history = model.fit_generator(data_generator(),
len(X_train) // batch_size,
validation_data=(X_test, y_test),
callbacks=cbks,
epochs=20)
loss = history.history['loss']
assert len(loss) > 1
assert loss[-1] == np.inf or np.isnan(loss[-1])
values = []
with open(fp) as f:
for x in reader(f):
values.append(x)
assert 'nan' in values[-1], 'The last epoch was not logged.'
os.remove(fp)
def train_a_model(trainfile):
'''
:param trainfile:
:return:
'''
#load the dataset using pandas
data = pd.read_csv(trainfile)
#check len of rows and its intance len
with open(trainfile) as f:
content = f.readlines()
lines = np.array(content)
num_of_instances = lines.size
print("number of instances: ", num_of_instances)
#save the data in train and validation data
x_train, y_train = [], []
for i in range(1, num_of_instances):
try:
emotion = lines[i].split(",")[0]
val = lines[i].split(",")[1:]
pixels = np.array(val, 'float32')
y_train.append(emotion)
x_train.append(pixels)
except:
print("", end="")
# data transformation for train and test sets
x_train = np.array(x_train, 'float32')
x_train /= 255 #normalize inputs between [0, 1]
x_train = x_train.reshape(x_train.shape[0], 48, 48, 1)
x_train = x_train.astype('float32')
print(x_train.shape[0], 'train samples')
y_train = []
for i in range(1, num_of_instances):
try:
emotion = lines[i].split(",")[0]
y_train.append(emotion)
except:
print("", end="")
le = LabelEncoder()
y_train = le.fit_transform(y_train)
a_train = []
for i in range(1, num_of_instances):
emotion = y_train[i - 1]
emotion = keras.utils.to_categorical(emotion, 3)
a_train.append(emotion)
y_train = a_train
y_train = np.array(y_train, 'float32')
#variables
num_classes = 3
#fear, happy, sad
batch_size = 256
epochs = 10
#construct CNN structure
model = Sequential()
#1st convolution layer
model.add(Conv2D(64, (5, 5), activation='relu', input_shape=(48, 48, 1)))
model.add(MaxPooling2D(pool_size=(5, 5), strides=(2, 2)))
#2nd convolution layer
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(AveragePooling2D(pool_size=(3, 3), strides=(2, 2)))
#3rd convolution layer
model.add(Conv2D(128, (3, 3), activation='relu'))
model.add(Conv2D(128, (3, 3), activation='relu'))
model.add(AveragePooling2D(pool_size=(3, 3), strides=(2, 2)))
model.add(Flatten())
#fully connected neural networks
model.add(Dense(1024, activation='relu'))
model.add(Dropout(0.2))
#-----------------------------
model.add(Dense(128))
model.add(Activation('relu'))
model.add(Dropout(0.3))
#-----------------------------
model.add(Dense(1024, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(num_classes, activation='softmax'))
################################
#------------------------------
#batch process
gen = ImageDataGenerator()
train_generator = gen.flow(x_train, y_train, batch_size=batch_size)
#------------------------------
model.compile(loss='categorical_crossentropy',
optimizer=keras.optimizers.Adam(),
metrics=['accuracy'])
#------------------------------
print('Training start please wait....')
#model.fit_generator(x_train, y_train, epochs=epochs) #train for all trainset
model.fit_generator(train_generator,
steps_per_epoch=batch_size,
epochs=epochs) #train for randomly selected one
print("model Training completed")
model.save('models.h5')
return model
#adam = keras.optimizers.Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.0, amsgrad=False)
#adagrad = keras.optimizers.Adagrad(lr=0.01, epsilon=None, decay=0.0)
# chose categorical_crossentropy to avoid slow learning for the sigma derivative.
model.compile(loss='categorical_crossentropy',
optimizer=keras.optimizers.Adadelta(),
metrics=['accuracy'])
#Create the generators for the test and train batches
training_gen = generate(batch_size, True)
valid_gen = generate(batch_size, False)
#fit the model
history = model.fit_generator(generator=training_gen,
verbose=1,
epochs=epochs,
steps_per_epoch=8000 / batch_size,
validation_data=valid_gen,
validation_steps=2000 / batch_size)
#score = model.evaluate(x_test, y_test, verbose=0)
#print('Test loss:', score[0])
#print('Test accuracy:', score[1])
model.save('./chest_xray_net.h5')
print(history.history.keys())
# summarize history for accuracy
plt.plot(history.history['acc'])
plt.plot(history.history['val_acc'])
plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
restnet.trainable = True
set_trainable = False
for layer in restnet.layers:
if layer.name in ['res5c_branch2b', 'res5c_branch2c', 'activation_49']:
set_trainable = True
if set_trainable:
layer.trainable = True
else:
layer.trainable = False
layers = [(layer, layer.name, layer.trainable) for layer in restnet.layers]
pd.DataFrame(layers, columns=['Layer Type', 'Layer Name', 'Layer Trainable'])
model_finetuned = Sequential()
model_finetuned.add(restnet)
model_finetuned.add(Dense(512, activation='relu', input_dim=(300, 300, 3)))
model_finetuned.add(Dropout(0.3))
model_finetuned.add(Dense(512, activation='relu'))
model_finetuned.add(Dropout(0.3))
model_finetuned.add(Dense(1, activation='sigmoid'))
model_finetuned.compile(loss='binary_crossentropy',
optimizer=optimizers.RMSprop(lr=1e-5),
metrics=['accuracy'])
model_finetuned.summary()
history_1 = model_finetuned.fit_generator(train_generator,
steps_per_epoch=100,
epochs=2,
validation_data=val_generator,
validation_steps=100,
verbose=1)
model_finetuned.save('men_women_tlearn_finetune_img_aug_restnet50.h5')
print("video: training ", model_name)
weights_path = "models_video/" + model_name + "_" + "{}".format(
deformation) + ".hdf5"
# base_model = cnnModels(model_name, image_shape=(64, 64, 1))
# print(base_model.model.summary())
load_weights = False
train_generator, validation_generator, steps_per_epoch = get_generators()
# Helper: TensorBoard
tb = TensorBoard(log_dir='./cnn_video/logs/' + '{}'.format(deformation))
# Helper: Stop when we stop learning.
early_stopper = EarlyStopping(patience=10)
# # Helper: Save results.
# timestamp = time.time()
# csv_logger = CSVLogger('./cnn_video/logs/' + model_name + '-' + 'training-' + \
# str(timestamp) + '.log')
model.fit_generator(train_generator,
steps_per_epoch=steps_per_epoch,
validation_data=validation_generator,
validation_steps=10,
callbacks=[tb, early_stopper],
epochs=epochs)
model.save_weights(weights_path, overwrite=True)
class Ghouzam(DigitClassifier):
'''Ghouzam digit classifier. Source : https://www.kaggle.com/yassineghouzam/introduction-to-cnn-keras-0-997-top-6?fbclid=IwAR12dON7-HGwZrjwan9H8UUaftW3Jk7nyv6mJuQ6keSr9yPEYEMwsQhQLdg'''
def __init__(self, warm_start=None, epochs=10, batch_size=86):
super().__init__(warm_start)
self.epochs = epochs
self.batch_size = batch_size
if self.model is None:
self.model = Sequential()
self.model.add(
Conv2D(filters=32,
kernel_size=(5, 5),
padding='Same',
activation='relu',
input_shape=(28, 28, 1)))
self.model.add(
Conv2D(filters=32,
kernel_size=(5, 5),
padding='Same',
activation='relu'))
self.model.add(MaxPool2D(pool_size=(2, 2)))
self.model.add(Dropout(0.25))
self.model.add(
Conv2D(filters=64,
kernel_size=(3, 3),
padding='Same',
activation='relu'))
self.model.add(
Conv2D(filters=64,
kernel_size=(3, 3),
padding='Same',
activation='relu'))
self.model.add(MaxPool2D(pool_size=(2, 2), strides=(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(10, activation='softmax'))
# Optimizer
optimizer = RMSprop(lr=0.001, rho=0.9, epsilon=1e-08, decay=0.0)
# Compile model
self.model.compile(optimizer=optimizer,
loss='categorical_crossentropy',
metrics=['accuracy'])
# Data augmentation to prevent overfitting
self.datagen = ImageDataGenerator(
featurewise_center=False, # set input mean to 0 over the dataset
samplewise_center=False, # set each sample mean to 0
featurewise_std_normalization=
False, # divide inputs by std of the dataset
samplewise_std_normalization=
False, # divide each input by its std
zca_whitening=False, # apply ZCA whitening
rotation_range=
10, # randomly rotate images in the range (degrees, 0 to 180)
zoom_range=0.1, # Randomly zoom image
width_shift_range=
0.1, # randomly shift images horizontally (fraction of total width)
height_shift_range=
0.1, # randomly shift images vertically (fraction of total height)
horizontal_flip=False, # randomly flip images
vertical_flip=False # randomly flip images
)
def fit(self, X_LS, y_LS, X_VS, y_VS):
'''Fits model.'''
# One-hot encoding
y_LS = to_categorical(y_LS)
y_VS = to_categorical(y_VS)
# Fit
self.datagen.fit(X_LS)
learning_rate_reduction = ReduceLROnPlateau(monitor='val_accuracy',
patience=3,
verbose=1,
factor=0.5,
min_lr=0.00001)
self.model.fit_generator(self.datagen.flow(X_LS,
y_LS,
batch_size=self.batch_size),
epochs=self.epochs,
validation_data=(X_VS, y_VS),
verbose=2,
steps_per_epoch=X_LS.shape[0] //
self.batch_size,
callbacks=[learning_rate_reduction])
return self
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True)
# this is the augmentation configuration we will use for testing:
# only rescaling
test_datagen = ImageDataGenerator(rescale=1. / 255)
train_generator = train_datagen.flow_from_directory(train_data_dir,
target_size=(img_width,
img_height),
batch_size=batch_size,
class_mode='binary')
validation_generator = test_datagen.flow_from_directory(
validation_data_dir,
target_size=(img_width, img_height),
batch_size=batch_size,
class_mode='binary')
#model.load_weights(networkFile)
history = model.fit_generator(train_generator,
steps_per_epoch=nb_train_samples // batch_size,
epochs=epochs,
validation_data=validation_generator,
validation_steps=nb_validation_samples //
batch_size)
printHistory(history, insulators)
model.save_weights(networkFile)
print(model.summary())
# ### Learning 1: New layers
epochs = 10
workers = 4
use_multiprocessing = False
print('Training for', epochs, 'epochs with', workers,
'workers, use_multiprocessing is', use_multiprocessing)
history = model.fit_generator(train_generator,
steps_per_epoch=nimages_train // batch_size,
epochs=epochs,
validation_data=validation_generator,
validation_steps=nimages_validation //
batch_size,
verbose=2,
callbacks=callbacks,
use_multiprocessing=use_multiprocessing,
workers=workers)
fname = "gtsrb-vgg16-reuse.h5"
print('Saving model to', fname)
model.save(fname)
# ### Learning 2: Fine-tuning
#
# Once the top layers have learned some reasonable weights, we can
# continue training by unfreezing the last convolution block of VGG16
# (`block5`) so that it may adapt to our data. The learning rate
# should be smaller than usual.
from keras.models import Sequential
from keras import layers
from keras.optimizers import RMSprop
from learning_keras.text_and_sequence.weather_data import float_data, train_gen, val_gen, val_steps
import matplotlib.pyplot as plt
model = Sequential()
model.add(
layers.GRU(32,
dropout=0.2,
recurrent_dropout=0.2,
input_shape=(None, float_data.shape[-1])))
model.add(layers.Dense(1))
model.compile(optimizer=RMSprop(), loss='mae')
history = model.fit_generator(train_gen,
steps_per_epoch=500,
epochs=40,
validation_data=val_gen,
validation_steps=val_steps)
loss = history.history['loss']
val_loss = history.history['val_loss']
epochs = range(1, len(loss) + 1)
plt.figure()
plt.plot(epochs, loss, 'bo', label='Training loss')
plt.plot(epochs, val_loss, 'b', label='Validation loss')
plt.title('Training and validation loss')
plt.legend()
plt.show()
def main(_):
##
# Load Data
##
with open(FLAGS.data_path, 'r') as f:
reader = csv.reader(f)
# data is a list of tuples (img path, steering angle)
data = np.array([row for row in reader])
# Split train and validation data
np.random.shuffle(data)
split_i = int(len(data) * 0.9)
X_train, y_train = list(zip(*data[:split_i]))
X_val, y_val = list(zip(*data[split_i:]))
X_train, y_train = np.array(X_train), np.array(y_train)
X_val, y_val = np.array(X_val), np.array(y_val)
##
# Define Model
##
model = Sequential([
Conv2D(32, (3, 3),
input_shape=(32, 128, 3),
padding='same',
activation='relu'),
Conv2D(64, (3, 3), padding='same', activation='relu'),
MaxPooling2D(),
Dropout(0.5),
Conv2D(128, (3, 3), padding='same', activation='relu'),
Conv2D(128, (3, 3), padding='same', activation='relu'),
MaxPooling2D(),
Dropout(0.5),
# Conv2D(1024, (3, 3), padding='same', activation='relu'),
# Conv2D(1024, (3, 3), padding='same', activation='relu'),
# MaxPooling2D(),
# Dropout(0.5),
# Conv2D(2048, (3, 3), padding='same', activation='relu'),
# Conv2D(2048, (3, 3), padding='same', activation='relu'),
# MaxPooling2D(),
# Dropout(0.5),
Flatten(),
Dense(1024, activation='relu'),
Dense(512, activation='relu'),
Dense(128, activation='relu'),
Dense(1, name='output', activation='tanh'),
])
model.compile(optimizer=Adam(lr=FLAGS.lrate), loss='mse')
##
# Train
##
history = model.fit_generator(
gen_batches(X_train, y_train, FLAGS.batch_size),
len(X_train) / FLAGS.batch_size,
FLAGS.num_epochs,
validation_data=gen_batches(X_val, y_val, FLAGS.batch_size),
validation_steps=(len(X_val) / FLAGS.batch_size))
##
# Save model
##
if not os.path.exists(FLAGS.save_dir):
os.makedirs(FLAGS.save_dir)
json = model.to_json()
model.save_weights(os.path.join(FLAGS.save_dir, 'model.h5'))
with open(os.path.join(FLAGS.save_dir, 'model.json'), 'w') as f:
f.write(json)
# compile and train the model using the generator function
train_generator = generator(train_samples, batch_size=batch)
validation_generator = generator(validation_samples, batch_size=batch)
model = Sequential()
model.add(Lambda(lambda x: (x / 255.0) - 0.5, input_shape=(160, 320, 3)))
model.add(Cropping2D(cropping=((60, 20), (0, 0))))
model.add(Conv2D(24, (5, 5), strides=(2, 2), activation="relu"))
model.add(Conv2D(36, (5, 5), strides=(2, 2), activation="relu"))
model.add(Conv2D(48, (5, 5), strides=(2, 2), activation="relu"))
model.add(Conv2D(64, (3, 3), activation="relu"))
model.add(Conv2D(64, (3, 3), activation="relu"))
model.add(Flatten())
model.add(Dense(100))
model.add(Dropout(0.5))
model.add(Dense(50))
model.add(Dropout(0.5))
model.add(Dense(10))
model.add(Dropout(0.3))
model.add(Dense(1))
model.compile(loss='mse', optimizer='adam')
model.fit_generator(train_generator,
steps_per_epoch=np.ceil(len(train_samples) / batch),
validation_data=validation_generator,
validation_steps=np.ceil(len(validation_samples) / batch),
epochs=5)
model.save('model.h5')
class Model():
def __init__(self):
self.model = Sequential()
def build_model(self, configs):
timer = Timer()
timer.start()
for layer in configs['model']['layers']:
neurons = layer['neurons'] if 'neurons' in layer else None
dropout_rate = layer['rate'] if 'rate' in layer else None
activation = layer['activation'] if 'activation' in layer else None
return_seq = layer['return_seq'] if 'return_seq' in layer else None
input_timesteps = layer[
'input_timesteps'] if 'input_timesteps' in layer else None
input_dim = layer['input_dim'] if 'input_dim' in layer else None
if layer['type'] == 'dense':
self.model.add(Dense(neurons, activation=activation))
if layer['type'] == 'lstm':
self.model.add(
CuDNNLSTM(neurons,
input_shape=(input_timesteps, input_dim),
return_sequences=return_seq))
if layer['type'] == 'dropout':
self.model.add(Dropout(dropout_rate))
if layer['type'] == 'BatchNormalization':
self.model.add(BatchNormalization())
self.model.compile(loss=configs['model']['loss'],
optimizer=configs['model']['optimizer'],
metrics=['accuracy'])
print('MODEL Compiled')
timer.stop()
def train(self, X, y, epochs, batch_size, save_dir, logs):
timer = Timer()
timer.start()
print('MODEL Training Started')
print(f'MODEL {epochs} epochs, {batch_size} batch size')
save_fname = os.path.join(
save_dir,
f'{dt.datetime.now().strftime("%d%m%Y-%H%M%S")}-e{epochs}.h5')
callbacks = [
EarlyStopping(monitor='val_loss', patience=2),
ModelCheckpoint(filepath=save_fname,
monitor='val_loss',
save_best_only=True),
TensorBoard(
log_dir=
f'{logs}/{dt.datetime.now().strftime("%d%m%Y-%H%M%S")}-e{epochs}'
)
]
self.model.fit(X,
y,
epochs=epochs,
batch_size=batch_size,
callbacks=callbacks,
shuffle=False)
self.model.save(save_fname)
print(f'MODEL Training Completed. Model saved as {save_fname}')
timer.stop()
def train_generator(self, data_gen, epochs, batch_size, steps_per_epoch,
save_dir):
timer = Timer()
timer.start()
print('MODEL Out-of-Memory Training Started')
print(
f'MODEL {epochs} epochs, {batch_size} batch size, {steps_per_epoch} batches per epoch'
)
save_fname = os.path.join(
save_dir, '%s-e%s.h5' %
(dt.datetime.now().strftime('%d%m%Y-%H%M%S'), str(epochs)))
callbacks = [
ModelCheckpoint(filepath=save_fname,
monitor='loss',
save_best_only=True)
]
self.model.fit_generator(data_gen,
steps_per_epoch=steps_per_epoch,
epochs=epochs,
callbacks=callbacks,
workers=1)
print(
f'MODEL Out-of-Memory Training Completed. Model saved as {save_fname}'
)
timer.stop()
def predict(self, x):
print('MODEL Predicting Point-by-Point:')
predicted = self.model.predict(x)
predicted = np.reshape(predicted, (predicted.size, ))
return predicted
def predict_sequences_multiple(self, data, window_size, prediction_len):
print('MODEL Predicting Sequences Multiple:')
prediction_seqs = []
for i in range(int(len(data) / prediction_len)):
curr_frame = data[i * prediction_len]
predicted = []
for j in range(prediction_len):
predicted.append(
self.model.predict(curr_frame[np.newaxis, :, :])[0, 0])
curr_frame = curr_frame[1:]
curr_frame = np.insert(curr_frame, [window_size - 2],
predicted[-1],
axis=0)
prediction_seqs.append(predicted)
return prediction_seqs
def evaluate(self, x, y):
return self.model.evaluate(x, y, verbose=0)
layer.trainable = False
model = Sequential()
model.add(Flatten(input_shape=base_model.output_shape[1:]))
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(5, activation='sigmoid'))
model = Model(inputs=base_model.input,
outputs=model(base_model.output)) # 新网络=预训练网络+自定义网络
model.compile(loss='binary_crossentropy',
optimizer=optimizers.SGD(lr=0.0001, momentum=0.9),
metrics=['accuracy'])
"""
train_datagen = ImageDataGenerator(rescale=1. / 255, horizontal_flip=True) # 训练数据预处理器，随机水平翻转
test_datagen = ImageDataGenerator(rescale=1. / 255) # 测试数据预处理器
train_generator = train_datagen.flow_from_directory(train_data_dir, target_size=(img_height, img_width),
batch_size=batch_size, class_mode='binary') # 训练数据生成器
validation_generator = test_datagen.flow_from_directory(validation_data_dir, target_size=(img_height, img_width),
batch_size=batch_size, class_mode='binary',
shuffle=False) # 验证数据生成器
checkpointer = ModelCheckpoint(filepath='dogcatmodel.h5', verbose=1, save_best_only=True) # 保存最优模型
# 训练&评估
model.fit_generator(train_generator, steps_per_epoch=nb_train_samples // batch_size, epochs=epochs,
validation_data=validation_generator, validation_steps=nb_validation_samples // batch_size,
verbose=2, workers=12, callbacks=[checkpointer])
"""
test_loss, test_acc = model.evaluate(x_test, y_test)
yield np.array(images), np.array(steer_angles)
model = Sequential()
model.add(Cropping2D(cropping=((70, 25), (0, 0)), input_shape=INPUT_SHAPE))
model.add(Lambda(lambda var: var / 127.5 - 1.0))
model.add(Conv2D(24, 5, 5, activation='relu', subsample=(2, 2)))
model.add(Conv2D(36, 5, 5, activation='relu', subsample=(2, 2)))
model.add(Conv2D(48, 5, 5, activation='relu', subsample=(2, 2)))
model.add(Conv2D(64, 3, 3, activation='relu'))
model.add(Conv2D(64, 3, 3, activation='relu'))
model.add(Dropout(KEEP_PROB))
model.add(Flatten())
model.add(Dense(100, activation='relu'))
model.add(Dense(50, activation='relu'))
model.add(Dense(10, activation='relu'))
model.add(Dense(1))
#model.summary()
model.compile(loss='mean_squared_error', optimizer=Adam(lr=LEARNING_RATE))
model.fit_generator(image_loader(x_train, y_train, BATCH_SIZE, True),
len(x_train),
EPOCHS,
max_q_size=1,
validation_data=image_loader(x_valid, y_valid, BATCH_SIZE,
False),
nb_val_samples=len(x_valid))
model.save('model.h5')
path = os.getcwd()
# this is the augmentation configuration we will use for training
train_datagen = ImageDataGenerator(
rescale=1./255)
#width_shift_range=0.1,
#height_shift_range=0.1,
#zoom_range=0.2,
#horizontal_flip=True)
train_generator = train_datagen.flow_from_directory(
path+'/Train', # this is the target directory
target_size=(256,256), # all images will be resized to 150x150
batch_size=batch_size,
class_mode='categorical') # since we use binary_crossentropy loss, we need binary labels
val_datagen = ImageDataGenerator(
rescale=1./255)
val_generator = val_datagen.flow_from_directory(
path+'/Test', # this is the target directory
target_size=(256,256), # all images will be resized to 150x150
batch_size=batch_size,
class_mode='categorical') # since we use binary_crossentropy loss, we need binary labels
model_3a.fit_generator(
train_generator,
steps_per_epoch=2000 // batch_size,validation_data=val_generator,validation_steps=100,
epochs=50,callbacks=callbacks)
# confirm the iterator works
#batchX, batchy = train_generator.next()
#print('Batch shape=%s, min=%.3f, max=%.3f' % (batchX.shape, batchX.min(), batchX.max()))
#batchX, batchy = validation_generator.next()
#print('Batch shape=%s, min=%.3f, max=%.3f' % (batchX.shape, batchX.min(), batchX.max()))
#batchX, batchy = test_generator.next()
#print('Batch shape=%s, min=%.3f, max=%.3f' % (batchX.shape, batchX.min(), batchX.max()))
# simple early stopping
es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=30)
mc = ModelCheckpoint('best_model.h5', monitor='val_acc', mode='max', verbose=1, save_best_only=True)
model.fit_generator(train_generator,
steps_per_epoch = train_generator.samples // batch_size, callbacks=[es, mc],
epochs = epochs, validation_data = validation_generator,
validation_steps = validation_generator.samples // batch_size)
"""
epochs = epochs, validation_data = validation_generator,
validation_steps = nb_validation_samples // batch_size)
"""
# load the saved model
saved_model = load_model('best_model.h5')
# make a prediction
test_generator.reset()
yhat = saved_model.predict_generator(test_generator, steps= nb_test_samples)
print(len(yhat))
#model.save_weights('model_saved.h5')
class Model:
def __init__(self):
self.model = None
self.history = LossHistory()
# モデリング
def build_model(self, dataset, nb_classes=3):
# 空のネットワークモデルを構築します。これは線形スタックモデルであり、各ニューラルネットワークレイヤーが順次追加されます。専門家の名前は順次モデルまたは線形スタックモデルです。
self.model = Sequential()
#次のコードは、CNNネットワークに必要なレイヤーを順次追加します。追加はネットワークレイヤーです
self.model.add(
Convolution2D(
32, 3, 3, border_mode='same',
input_shape=dataset.input_shape)) #レイヤー1：2次元たたみ込みレイヤー
self.model.add(Activation('relu')) #レイヤー1：アクティベーション機能
self.model.add(Convolution2D(32, 3, 3)) #レイヤー2：2次元畳み込みレイヤー
self.model.add(Activation('relu')) #レイヤー2：アクティベーション機能
#プーリング層の役割：
#1.invariance:translation,rotation,scale
#2.主な特徴を維持しながら、次元の削減が行われ、過剰適合を防止し、モデルの汎化能力を改善します
self.model.add(MaxPooling2D(pool_size=(2, 2))) #レイヤー3：プーリング層
self.model.add(
Dropout(0.25)) #レイヤー3：Dropout——このレイヤーの各ノードは、非アクティブ化の確率が25％です。
self.model.add(Convolution2D(64, 3, 3,
border_mode='same')) #レイヤー4:2次元たたみ込みレイヤー
self.model.add(Activation('relu')) #レイヤー4:アクティベーション機能
self.model.add(Convolution2D(64, 3, 3)) #レイヤー5:2维卷积层
self.model.add(Activation('relu')) #レイヤー5:アクティベーション機能
self.model.add(MaxPooling2D(pool_size=(2, 2))) #レイヤー6：プーリング層
self.model.add(Dropout(0.25)) #レイヤー6:Dropout
self.model.add(Flatten()) #レイヤー7:Flatten()——多次元入力1次元になる
self.model.add(Dense(512)) #レイヤー7:Dense層,完全に接続されたレイヤー
self.model.add(Activation('relu')) #レイヤー7:アクティベーション機能
self.model.add(Dropout(0.5)) #レイヤー7:Dropout
self.model.add(Dense(nb_classes)) #レイヤー８:Dense層
self.model.add(Activation('softmax')) #レイヤー８:分類層、最終結果を出力
#出力モデルの概要
self.model.summary()
# トレーニングモデル
def train(self,
train_images,
train_labels,
valid_images,
valid_labels,
batch_size=20,
nb_epoch=10,
data_augmentation=False):
sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True
) # SGD + momentumのオプティマイザをトレーニングに使用し、最初にオプティマイザオブジェクトを生成します
self.model.compile(loss='categorical_crossentropy',
optimizer=sgd,
metrics=['accuracy']) # 実際のモデル構成作業を完成する
# データ拡張を使用しない場合、いわゆる拡張とは、回転、反転、ノイズの追加などによって提供するトレーニングデータから新しいトレーニングデータを作成し、トレーニングデータのサイズを意識的に増やし、モデルトレーニングの量を増やすことです。
if not data_augmentation:
self.model.fit(train_images,
train_labels,
batch_size=batch_size,
nb_epoch=nb_epoch,
validation_data=(valid_images, valid_labels),
shuffle=True,
callbacks=[self.history])
#リアルタイムデータを使用して改善する
else:
# データプロモーション用のデータジェネレーターを定義します。ジェネレーターオブジェクトdatagenを返します。datagenが呼び出されるたびに一連のデータを生成し（順次生成）、メモリを節約します。これは実際にはpythonデータジェネレーターです
datagen = ImageDataGenerator(
featurewise_center=False, # 入力データを分散化するかどうか（平均値は0）
samplewise_center=False, # 入力データの各サンプルを0にするかどうか
featurewise_std_normalization=
False, # データが標準化されているかどうか（データセットの標準偏差で割った入力データ）
samplewise_std_normalization=False, # 各サンプルデータを独自の標準偏差で除算するかどうか
zca_whitening=False, # 入力データにZCAホワイトニングを適用するかどうか
rotation_range=20, # データが増加したときの画像のランダムな回転の角度（範囲0〜180）
width_shift_range=
0.2, # データがプロモートされたときの画像の水平オフセットの振幅（単位は画像の幅の割合、0〜1の浮動小数点数）
height_shift_range=0.2, # 上記と同じですが、ここは垂直です
horizontal_flip=True, # ランダムな水平反転を実行するかどうか
vertical_flip=False) # ランダム垂直フリップを実行するかどうか
# 固有値の正規化、ZCAホワイトニングなどのトレーニングサンプルセット全体の数を計算します。
datagen.fit(train_images)
# ジェネレーターを使用してモデルのトレーニングを開始します
self.model.fit_generator(datagen.flow(train_images,
train_labels,
batch_size=batch_size),
samples_per_epoch=train_images.shape[0],
nb_epoch=nb_epoch,
validation_data=(valid_images,
valid_labels),
callbacks=[self.history])
MODEL_PATH = os_path + '/data/model/aggregate.face.model.h5'
def save_model(self, file_path=MODEL_PATH):
self.model.save(file_path)
def load_model(self, file_path=MODEL_PATH):
self.model = load_model(file_path)
def evaluate(self, test_images, test_labels):
score = self.model.evaluate(test_images, test_labels, verbose=1)
print("%s: %.2f%%" % (self.model.metrics_names[1], score[1] * 100))
#ch, row, col = 3, 80, 320 # Trimmed image format
model = Sequential()
model.add(Lambda(lambda x: (x / 127.5) - 1., input_shape=(160, 320, 3)))
model.add(Cropping2D(cropping=((70, 25), (0, 0))))
model.add(Conv2D(24, (5, 5), strides=(2, 2), activation='relu'))
model.add(Conv2D(36, (5, 5), strides=(2, 2), activation='relu'))
model.add(Conv2D(48, (5, 5), strides=(2, 2), activation='relu'))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(Flatten())
model.add(Dense(100))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(50))
model.add(Activation('relu'))
model.add(Dense(10))
model.add(Activation('relu'))
model.add(Dense(1))
model.compile(optimizer='adam', loss='mse')
model.fit_generator(
train_generator,
steps_per_epoch=math.ceil(len(train_samples) / batch_size * 6),
validation_data=validation_generator,
validation_steps=math.ceil(len(validation_samples) / batch_size * 6),
epochs=3,
verbose=1)
#model.fit(X_train, y_train, validation_split = 0.2, shuffle = True, nb_epoch = 3)
model.save('model.h5')
test_datagen = ImageDataGenerator(rescale=1. / 255) # Preprocess images
training_set = train_datagen.flow_from_directory('dataset/training_set',
target_size=(64, 64),
batch_size=32,
class_mode='binary')
test_set = test_datagen.flow_from_directory('dataset/test_set',
target_size=(64, 64),
batch_size=32,
class_mode='binary')
classifier.fit_generator(training_set,
steps_per_epoch=8000 / 32,
epochs=25,
validation_data=test_set,
validation_steps=2000 / 32)
from keras.preprocessing import image
import numpy as np
# Single image prediction
test_img = image.load_img('dataset/single_prediction/cat_or_dog_1.jpg',
target_size=(64, 64))
test_img = image.img_to_array(test_img)
test_img = np.expand_dims(test_img, axis=0) # Preprocess image
result = classifier.predict(test_img)
training_set.class_indices # show indices for classification of cat / dog
# Convert binary result to String prediction result
model.add(Conv2D(32, (3, 3), padding='same'))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(64, (3, 3), padding='same'))
model.add(Activation('relu'))
model.add(Conv2D(64, 3, 3))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(512))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(10))
model.add(Activation('softmax'))
model.summary()
# compile
model.compile(loss='categorical_crossentropy', optimizer=OPTIM,
metrics=['accuracy'])
# train
history = model.fit_generator(datagen.flow(X_train, y_train, batch_size=BATCH_SIZE),
validation_data=(X_test, y_test),
steps_per_epoch=X_train.shape[0] // BATCH_SIZE,
epochs=NB_EPOCH, verbose=VERBOSE)
# evaluate
score = model.evaluate(X_test, y_test,
batch_size=BATCH_SIZE, verbose=VERBOSE)
print("Test score:", score[0])
print('Test accuracy:', score[1])
rescale=None,
# set function that will be applied on each input
preprocessing_function=None,
# image data format, either "channels_first" or "channels_last"
data_format=None,
# fraction of images reserved for validation (strictly between 0 and 1)
validation_split=0.0)
# Compute quantities required for feature-wise normalization
# (std, mean, and principal components if ZCA whitening is applied).
datagen.fit(x_train)
# Fit the model on the batches generated by datagen.flow().
model.fit_generator(datagen.flow(x_train, y_train,
batch_size=batch_size),
epochs=epochs,
steps_per_epoch=len(x_train)/batch_size*2,
validation_data=(x_test, y_test),
workers=4)
# Save model and weights
if not os.path.isdir(save_dir):
os.makedirs(save_dir)
model_path = os.path.join(save_dir, model_name)
model.save(model_path)
print('Saved trained model at %s ' % model_path)
# Score trained model.
scores = model.evaluate(x_test, y_test, verbose=1)
print('Test loss:', scores[0])
print('Test accuracy:', scores[1])
#model.add(Dropout(0.4))
model.add(Dense(units=1024))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(units=1024))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(units=512))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(units=7))
model.add(Activation('softmax'))
model.summary()
model.compile(optimizer="Adamax",loss='categorical_crossentropy',metrics=['accuracy'])
ep=689
model.fit_generator(datagen.flow(x,y, batch_size=batch),steps_per_epoch=len(x)/batch, epochs=ep)
ev=model.evaluate(xv,yv,batch_size=batch)
print (' ',ev[1],'\n')
print ('save')
model.summary()
model.save('save.h5')
class OCR:
def __init__(self, test_dir=dir_path + '/train_images/'):
self.learning_rate = 0.001
self.test_dir = test_dir
self.data = []
def format_y(self, y):
res = [0] * 128
if y == None:
return res
for i in range(len(y)):
if y[i] == None:
continue
res[y[i] - 2304] = 1
return res
def get_character(self, text):
chars = []
split = text.split('_')
split = split[3:]
if len(split) <= 0:
print('Image label not found for ', text)
return
split[-1] = split[-1][:-4] #remove .png from last split
for c in range(len(split)):
chars.append(int(split[c]))
return chars
def load_data(self):
if not self.test_dir:
print(
"Test directory could not be found. Please specify the correct path."
)
return
images_name = [
f for f in listdir(self.test_dir) if isfile(join(self.test_dir, f))
]
X = []
y = []
for file_name in images_name:
file_path = self.test_dir
img = cv2.imread(file_path + file_name, 0)
img = self.preprocess(img)
X.append(img.reshape(64, 64, 1))
chars = self.get_character(file_name)
y.append(self.format_y(chars))
return np.array(X), np.array(y)
def preprocess(self, img):
cv2.normalize(img, img, 0, 255, cv2.NORM_MINMAX)
kernel = np.ones((5, 5), np.float32) / 25
img = cv2.filter2D(img, -1, kernel)
ret, ret_img = cv2.threshold(img, 0, 255,
cv2.THRESH_BINARY_INV | cv2.THRESH_OTSU)
resz_img = cv2.resize(ret_img, (64, 64))
kernel1 = np.ones((3, 3), np.uint8)
erosion = cv2.erode(resz_img, kernel1, iterations=1)
kernel2 = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3, 3))
closing = cv2.morphologyEx(erosion, cv2.MORPH_CLOSE, kernel2)
ret, res_img = cv2.threshold(resz_img, 0, 255, cv2.THRESH_BINARY_INV)
res_img = cv2.filter2D(res_img, -1, kernel)
return res_img
def make_model(self):
self.model = Sequential()
self.model.add(
Conv2D(32,
kernel_size=(3, 3),
strides=(1, 1),
activation='relu',
input_shape=(64, 64, 1)))
self.model.add(Dropout(0.2))
self.model.add(Conv2D(64, (3, 3), activation='relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2)))
self.model.add(Conv2D(128, (3, 3), activation='relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2)))
self.model.add(Conv2D(256, (3, 3), activation='relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2)))
self.model.add(Conv2D(512, (3, 3), activation='relu'))
self.model.add(MaxPooling2D(pool_size=(2, 2)))
self.model.add(Flatten())
self.model.add(Dropout(0.2))
self.model.add(Dense(1024, activation='relu'))
self.model.add(Dropout(0.2))
self.model.add(Dense(512, activation='relu'))
self.model.add(Dropout(0.2))
self.model.add(Dense(128, activation='sigmoid'))
self.model.compile(loss='binary_crossentropy',
optimizer=Adam(lr=self.learning_rate),
metrics=[categorical_accuracy])
def train(self):
X, Y = self.load_data(
) #loads train data, preprocess it and return in desired format
X = X.astype('float32')
train_datagen = ImageDataGenerator(rotation_range=20,
rescale=1. / 255,
width_shift_range=0.1,
height_shift_range=0.1,
shear_range=0.2,
zoom_range=0.1,
horizontal_flip=False,
fill_mode='nearest')
X_train, X_test, Y_train, Y_test = train_test_split(X,
Y,
test_size=0.3)
X_test, X_validate, Y_test, Y_validate = train_test_split(
X_test, Y_test, test_size=0.5)
X_test = X_test / 255.
train_generator = train_datagen.flow(X_train, Y_train, batch_size=32)
validation_generator = train_datagen.flow(X_validate,
Y_validate,
batch_size=32)
if isfile(join(dir_path, '/my_model_new.h5')):
self.model = load_model('my_model_new.h5')
else:
self.make_model()
self.model.fit_generator(train_generator,
steps_per_epoch=len(X_train) / 32,
epochs=50,
validation_data=validation_generator,
validation_steps=len(X_validate) / 32)
score, acc = self.model.evaluate(X_test, Y_test)
self.model.save('my_model_new.h5')
print('score: ', score, 'accuracy: ', acc)
'pred_prob': pred_probs,
}
for c, col in zip(class_cols, class_probs):
frame_dict[c] = col
table = pandas.DataFrame(frame_dict, columns=all_cols)
number_cols = ['true_prob', 'pred_prob'] + class_cols
table[number_cols] = table[number_cols].apply(pandas.to_numeric)
#from IPython import embed; embed()
return table
class ResultsDataFrameCallback(keras.callbacks.Callback):
def on_epoch_end(self, epoch, logs=None):
run.summary["results"] = results_data_frame(test_datagen, model)
model.fit_generator(train_generator,
steps_per_epoch=len(train_generator),
epochs=config.epochs,
workers=4,
callbacks=[ResultsDataFrameCallback()],
validation_data=test_generator,
validation_steps=len(test_generator))
if config.epochs == 0:
#run.summary["results"] = results_data_frame(test_datagen, model)
run.summary.update({"results2": results_data_frame(test_datagen, model)})
model.summary()
model.get_config()
model.layers[0].get_config()
model.layers[0].input_shape
model.layers[0].output_shape
model.layers[0].get_weights()
numpy.shape(model.layers[0].get_weights()[0])
model.layers[0].trainable
filepath = weights_filename
checkpoint = callbacks.ModelCheckpoint(filepath,
monitor='val_acc',
verbose=1,
save_best_only=True,
mode='max')
callbacks_list = [checkpoint]
# Training
hist = model.fit_generator(train_it,
steps_per_epoch=5703,
validation_data=val_it,
validation_steps=1140,
epochs=epochs,
callbacks=callbacks_list)
# Evaluating the model
score = model.evaluate_generator(test_it, steps=3421)
#score = model.evaluate(X_test, y_test, verbose=0)
print('Test Loss:', score[0])
print('Test accuracy:', score[1])
######################################################
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dense(2, activation='softmax'))
model.compile(loss=keras.losses.categorical_crossentropy,
optimizer=keras.optimizers.Adam(),
metrics=['accuracy'])
print(model.summary())
# callbacks = [LearningRateScheduler(poly_decay)]
history = model.fit_generator(trainGen,
steps_per_epoch=totalTrain // BS,
validation_data=valGen,
validation_steps=totalVal // BS,
epochs=NUM_EPOCHS,
verbose=1
# callbacks=callbacks
)
print("[INFO] evaluating model...")
testGen.reset()
# predIdxs = model.predict_generator(testGen,
# steps=(totalTest//BS) + 1)
# predIdxs = np.argmax(predIdxs, axis=1)
score = model.evaluate_generator(testGen, verbose=1)
model.save("saved_model.model")
class Model():
"""A class for an building and inferencing an lstm model"""
def __init__(self):
self.model = Sequential()
def load_model(self, filepath):
print('[Model] Loading model from file %s' % filepath)
self.model = load_model(filepath)
def build_model(self, configs):
timer = Timer()
timer.start()
for layer in configs['model']['layers']:
neurons = layer['neurons'] if 'neurons' in layer else None
dropout_rate = layer['rate'] if 'rate' in layer else None
activation = layer['activation'] if 'activation' in layer else None
return_seq = layer['return_seq'] if 'return_seq' in layer else None
input_timesteps = layer[
'input_timesteps'] if 'input_timesteps' in layer else None
input_dim = layer['input_dim'] if 'input_dim' in layer else None
if layer['type'] == 'dense':
self.model.add(Dense(neurons, activation=activation))
if layer['type'] == 'lstm':
self.model.add(
LSTM(neurons,
input_shape=(input_timesteps, input_dim),
return_sequences=return_seq))
if layer['type'] == 'dropout':
self.model.add(Dropout(dropout_rate))
self.model.compile(loss=configs['model']['loss'],
optimizer=configs['model']['optimizer'])
print('[Model] Model Compiled')
timer.stop()
def train(self, x, y, epochs, batch_size, save_dir):
timer = Timer()
timer.start()
print('[Model] Training Started')
print('[Model] %s epochs, %s batch size' % (epochs, batch_size))
save_fname = os.path.join(
save_dir, '%s-e%s.h5' %
(dt.datetime.now().strftime('%d%m%Y-%H%M%S'), str(epochs)))
callbacks = [
EarlyStopping(monitor='val_loss', patience=2),
ModelCheckpoint(filepath=save_fname,
monitor='val_loss',
save_best_only=True)
]
self.model.fit(x,
y,
epochs=epochs,
batch_size=batch_size,
callbacks=callbacks)
self.model.save(save_fname)
print('[Model] Training Completed. Model saved as %s' % save_fname)
timer.stop()
def train_generator(self, data_gen, epochs, batch_size, steps_per_epoch,
save_dir):
timer = Timer()
timer.start()
print('[Model] Training Started')
print('[Model] %s epochs, %s batch size, %s batches per epoch' %
(epochs, batch_size, steps_per_epoch))
save_fname = os.path.join(
save_dir, '%s-e%s.h5' %
(dt.datetime.now().strftime('%d%m%Y-%H%M%S'), str(epochs)))
callbacks = [
ModelCheckpoint(filepath=save_fname,
monitor='loss',
save_best_only=True)
]
self.model.fit_generator(data_gen,
steps_per_epoch=steps_per_epoch,
epochs=epochs,
callbacks=callbacks,
workers=1)
print('[Model] Training Completed. Model saved as %s' % save_fname)
timer.stop()
def predict_point_by_point(self, data):
#Predict each timestep given the last sequence of true data, in effect only predicting 1 step ahead each time
print('[Model] Predicting Point-by-Point...')
predicted = self.model.predict(data)
predicted = np.reshape(predicted, (predicted.size, ))
return predicted
def predict_sequences_multiple(self, data, window_size, prediction_len):
#Predict sequence of 50 steps before shifting prediction run forward by 50 steps
print('[Model] Predicting Sequences Multiple...')
prediction_seqs = []
for i in range(int(len(data) / prediction_len)):
curr_frame = data[i * prediction_len]
predicted = []
for j in range(prediction_len):
predicted.append(
self.model.predict(curr_frame[newaxis, :, :])[0, 0])
curr_frame = curr_frame[1:]
curr_frame = np.insert(curr_frame, [window_size - 2],
predicted[-1],
axis=0)
prediction_seqs.append(predicted)
return prediction_seqs
def predict_sequence_full(self, data, window_size, prediction_len):
#Shift the window by 1 new prediction each time, re-run predictions on new window
print('[Model] Predicting Sequences Full...')
curr_frame = data[0]
predicted = []
for i in range(len(data)):
predicted.append(
self.model.predict(curr_frame[newaxis, :, :])[0, 0])
curr_frame = curr_frame[1:]
curr_frame = np.insert(curr_frame, [window_size - 2],
predicted[-1],
axis=0)
return predicted
horizontal_flip=True)
# this is the augmentation configuration we will use for testing:
# only rescaling
test_datagen = ImageDataGenerator(rescale=1. / 255)
# this is a generator that will read pictures found in
# subfolers of 'data/train', and indefinitely generate
# batches of augmented image data
train_generator = train_datagen.flow_from_directory(
'data/train', # this is the target directory
target_size=(150, 150), # all images will be resized to 150x150
batch_size=batch_size,
class_mode='binary'
) # since we use binary_crossentropy loss, we need binary labels
# this is a similar generator, for validation data
validation_generator = test_datagen.flow_from_directory('data/validation',
target_size=(150, 150),
batch_size=batch_size,
class_mode='binary')
model.fit_generator(train_generator,
steps_per_epoch=2000 // batch_size,
epochs=50,
validation_data=validation_generator,
validation_steps=800 // batch_size)
model.save_weights(
'first_try.h5'
) # always save your weights after training or during training
nb_batches=nb_val_samples // batch_size,
batch_size=batch_size)
# In[67]:
r = next(traingen)
r[0].shape, r[1].shape, len(r)
# In[74]:
for iteration in range(1):
print 'Iteration', iteration
print "1"
h = model.fit_generator(traingen,
steps_per_epoch=nb_train_samples // batch_size,
epochs=1,
verbose=1,
validation_data=valgen,
validation_steps=nb_val_samples)
print "2"
for k, v in h.history.iteritems():
history[k] = history.get(k, []) + v
print "3"
with open('%s.history.pkl' % FN, 'wb') as fp:
pickle.dump(history, fp, -1)
print "4"
model.save_weights('%s.hdf5' % FN, overwrite=True)
print "5"
gensamples(batch_size=batch_size)
# In[ ]:
training_set = train_datagen.flow_from_directory(
'train',
target_size=(150, 150),
batch_size=16,
class_mode='binary')
test_set = test_datagen.flow_from_directory(
'test',
target_size=(150, 150),
batch_size=16,
class_mode='binary')
model_saved = model.fit_generator(
training_set,
epochs=10,
validation_data=test_set,
)
model.save('mymodel.h5', model_saved)
# To test for individual images
mymodel = load_model('mymodel.h5')
# test_image=image.load_img('C:/Users/Karan/Desktop/ML Datasets/Face Mask Detection/Dataset/test/without_mask/30.jpg',target_size=(150,150,3))
test_image = image.load_img(r'C:/Users/karan/Desktop/FaceMaskDetector/test/with_mask/1-with-mask.jpg',
target_size=(150, 150, 3))
test_image
test_image = image.img_to_array(test_image)
test_image = np.expand_dims(test_image, axis=0)
mymodel.predict(test_image)[0][0]
model.add(Dense(64, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(1001, activation='softmax'))
#model.add(Activation('softmax'))
model.summary()
model.compile(loss='sparse_categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
nb_epoch = 1
nb_train_samples = 63
nb_validation_samples = 55
model.fit_generator(train_generator,
samples_per_epoch=nb_train_samples,
nb_epoch=nb_epoch,
validation_data=validation_generator,
nb_val_samples=nb_validation_samples)
model.evaluate_generator(validation_generator, nb_validation_samples)
model.save('/home/garima/Music/dogs-vs-cats/models/fromsiraj.h5')
new_model = keras.models.load_model(
'/home/garima/Music/dogs-vs-cats/models/fromsiraj.h5')
converter = tf.lite.TocoConverter.from_keras_model_file(
'/home/garima/Music/dogs-vs-cats/models/fromsiraj.h5')
tflite_model = converter.convert()
open(
"/home/garima/Downloads/examples-master/lite/examples/image_classification/android/app/src/main/assets/from2.tflite",
"wb").write(tflite_model)
