def test_cnn(sess):
data_loader_test_data = dataset_loader.dataset(batch_size=batch_size, test_percentage=test_percentage,
validation_percentage=validation_percentage)
image_l, image_lab = data_loader_test_data.getTestData()
# image = cv2.imread(dirName + 'sample.JPEG')
# image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
# images = []
# images.append(image)
#
# images = np.array(images)
#
# image_l, images = data_loader.rgb2lab(images)
# images = tf.cast(images, tf.float32)
# # image_l = tf.cast(image_l, tf.float32)
image_l = np.array(image_l, dtype=np.float32)
# image_l = images[:,:,:,0:1]
encoded_img = convolutional_neural_network(image_l)
# sess.run(tf.global_variables_initializer())
encoded_img_val = sess.run(encoded_img)
# image_l = np.array(sess.run([image_l]))
images_rgb = decode_batch(image_l,encoded_img_val,2.63)
i = 0
for image_rgb in images_rgb:
i+=1
imsave(dirName + str(i) + 'Gen.jpeg', image_rgb)
i = 0
def test_encode(dirName="generatedPics/"):
data_loader = dataset_loader.dataset()
image = cv2.imread(dirName + 'sample.JPEG')
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
images = []
images.append(image)
images = np.array(images)
image_l, images = data_loader.rgb2lab(images)
# image_l = images[:,:,:,0:1]
# pprint.pprint(images.shape)
images = encode(images)
# pprint.pprint(images.shape)
image = decode(image_l, images, 2.63)
# image = decode(X_l, sess.run(prediction))
imsave(dirName + 'sample2.jpeg', image)
def train_neural_network(X):
x_ab_ss = X[:, ::4, ::4, :]
prediction = convolutional_neural_network(X)
thresh = 5
nongray_mask = (np.sum(np.sum(np.sum(np.abs(x_ab_ss) > thresh, axis=1), axis=1), axis=1) > 0)[:, np.newaxis, np.newaxis, np.newaxis]
#Prior_Boost
#prior_boost: [N, 1, H/4, W/4]
prior_boost = _prior_boost(prediction)
#Eltwise
#prior_boost_nongray: [N, 1, H/4, W/4]
prior_boost_nongray = prior_boost * nongray_mask
g_loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=prediction, labels=Y))
dl2c = tf.gradients(g_loss, prediction)
dl2c = tf.stop_gradient(dl2c)
#
cost = tf.reduce_sum(dl2c * prediction * prior_boost_nongray)
optimizer = tf.train.AdamOptimizer(learning_rate=0.000001).minimize(cost)
hm_epochs = 50
with tf.device("/gpu:0"):
with tf.Session(config=tf.ConfigProto(log_device_placement=False)) as sess:
sess.run(tf.global_variables_initializer())
for epoch in range(hm_epochs):
epoch_loss = 0
data_loader = dataset_loader.dataset(batch_size=batch_size, test_percentage=test_percentage,
validation_percentage=validation_percentage)
for _ in range(int(train_size / batch_size)):
epoch_x, epoch_y = data_loader.getNextBatch()
encoded_epoch_y = vectorized_encode(epoch_y)
_, c = sess.run([optimizer, cost], feed_dict={X: epoch_x, Y: encoded_epoch_y})
# print "Cost: ", c
epoch_loss += c
print('Epoch', epoch, 'completed out of', hm_epochs, 'loss:', epoch_loss)
test_cnn(sess)
# image_l = np.array(sess.run([image_l]))
images_rgb = decode_batch(image_l,encoded_img_val,2.63)
i = 0
for image_rgb in images_rgb:
i+=1
imsave(dirName + str(i) + 'Gen.jpeg', image_rgb)
i = 0
# for image_test in image_lab:
# i+=1
# imsave(dirName + str(i) + 'Test.jpeg', image_test)
batch_size = 3
test_percentage = 15
validation_percentage = 10
data_loader = dataset_loader.dataset(batch_size = batch_size, test_percentage = test_percentage, validation_percentage = validation_percentage)
train_size = data_loader.n_train_records
pprint = pprint.PrettyPrinter(indent=4)
dirName = "generatedPics/"
X = tf.placeholder(tf.float32,shape=[None,256,256,1])
Y = tf.placeholder(tf.float32,shape=[None,64,64,313])
def train_neural_network(X):
x_ab_ss = X[:, ::4, ::4, :]
prediction = convolutional_neural_network(X)
thresh = 5
nongray_mask = (np.sum(np.sum(np.sum(np.abs(x_ab_ss) > thresh, axis=1), axis=1), axis=1) > 0)[:, np.newaxis, np.newaxis, np.newaxis]
#Prior_Boost
#prior_boost: [N, 1, H/4, W/4]
def cnn_constructor():
"""
Referenced by https://github.com/oreilly-japan/deep-learning-from-scratch
common modules referenced there too.
"""
global network, classes, imsize
(x_train, t_train), (x_test,
t_test), classes = dataset(image_dir="images",
test_percentage=10,
validation_percentage=10,
imsize=imsize)
x_train = chenneling(x_train)
x_test = chenneling(x_test)
train_num = x_train.shape[0]
test_num = x_test.shape[0]
x_train, t_train = shuffle_dataset(x_train, t_train)
x_test, t_test = shuffle_dataset(x_test, t_test)
net_param = "cnn_params" + str(imsize) + ".pkl"
if not os.path.exists("params/"):
os.makedirs("params/")
# make convolution eural network
# x_train.shape[1:] returns channel, height, width
network = ConvNet(input_dim=(x_train.shape[1:]),
conv_param={
'filter_num': 20,
'filter_size': 3,
'pad': 0,
'stride': 1
},
hidden_size=32,
output_size=classes,
weight_init_std=0.001)
trainer = Trainer(network,
x_train,
t_train,
x_test,
t_test,
epochs=1,
mini_batch_size=FLAGS.batch_size,
optimizer='Adam',
optimizer_param={'lr': 0.001},
evaluate_sample_num_per_epoch=train_num)
params_loaded = False
if not os.path.exists("params/"):
os.makedirs("params/")
if (os.path.exists("params/" + net_param)):
network.load_params("params/" + net_param)
params_loaded = True
print("\n* Loaded Network Parameters! - " + net_param)
if ((FLAGS.train_epochs > 0) or (params_loaded == False)):
if (FLAGS.train_epochs <= 0):
FLAGS.train_epochs = 10
# Training
for ep in range(FLAGS.train_epochs):
trainer.train()
# Save parameters
network.save_params("params/" + net_param)
# plot graphs
# Grpah 1: Accuracy
markers = {'train': 'o', 'test': 's', 'loss': 'd'}
x1 = np.arange(len(trainer.train_acc_list))
plt.clf()
plt.plot(x1,
trainer.train_acc_list,
marker='o',
label='train',
markevery=1)
plt.plot(x1,
trainer.test_acc_list,
marker='s',
label='test',
markevery=1)
plt.xlabel("epochs")
plt.ylabel("accuracy")
plt.ylim(0, 1.1)
plt.legend(loc='lower right')
plt.title("Accuracy")
now = datetime.now()
filename = "params/" + now.strftime(
'%Y%m%d_%H%M%S%f') + "_" + "ep" + ".png"
plt.savefig(filename)
#plt.show()
# Graph 2: Loss
x2 = np.arange(len(trainer.train_loss_list))
plt.clf()
plt.plot(x2,
trainer.train_loss_list,
marker='o',
label='loss',
markevery=1)
plt.xlabel("iter")
plt.ylabel("loss")
plt.legend(loc='lower right')
plt.title("Cross entropy loss")
now = datetime.now()
filename = "params/" + now.strftime(
'%Y%m%d_%H%M%S%f') + "_" + "ep" + ".png"
plt.savefig(filename)
#plt.show()
print("\n* Saved Network Parameters! - " + net_param)