def store_prediction(self, sess, batch_x, batch_y, name):
prediction = sess.run(self.net.predicter,
feed_dict={
self.net.x: batch_x,
self.net.y: batch_y,
self.net.keep_prob: 1.,
self.net.block_size: 1
})
pred_shape = prediction.shape
loss = sess.run(self.net.cost,
feed_dict={
self.net.x: batch_x,
self.net.y:
util.crop_to_shape(batch_y, pred_shape),
self.net.keep_prob: 1.,
self.net.block_size: 1
})
logging.info("Verification error= {:.1f}%, loss= {:.4f}".format(
error_rate(prediction,
util.crop_to_shape(batch_y, prediction.shape)), loss))
img = util.combine_img_prediction(batch_x, batch_y, prediction)
util.save_image(img, "%s/%s.jpg" % (self.prediction_path, name))
return pred_shape
from tf_SDUnet import unet, util, image_util, metrics
import numpy as np
data_provider = image_util.ImageDataProvider("DRIVE700/test/*",
data_suffix="_test.tif",
mask_suffix='_manual1.png',
n_class=2)
net = unet.Unet(layers=4, features_root=8, channels=3, n_class=2)
test_x, test_y = data_provider(1)
prediction = net.predict("mode/drive100_0.92_700/model.ckpt", test_x)
prediction = util.crop_to_shape(prediction, (20, 584, 565, 2))
AUC_ROC = metrics.roc_Auc(prediction,
util.crop_to_shape(test_y, prediction.shape))
print("auc", AUC_ROC)
acc = metrics.acc(prediction, util.crop_to_shape(test_y, prediction.shape))
print("acc:", acc)
precision = metrics.precision(prediction,
util.crop_to_shape(test_y, prediction.shape))
print("ppv:", precision)
sen = metrics.sen(prediction, util.crop_to_shape(test_y, prediction.shape))
print("TPR:", sen)
TNR = metrics.TNR(prediction, util.crop_to_shape(test_y, prediction.shape))
print("tnr:", TNR)
f1 = metrics.f1score2(prediction, util.crop_to_shape(test_y, prediction.shape))
print("f1:", f1)
img = util.combine_img_prediction(test_x, test_y, prediction)
util.save_image(img, "19.jpg")
import os
data_provider2 = image_util.ImageDataProvider("TNBC/test/*.png",
data_suffix=".png",
mask_suffix='_mask.png',
n_class=2)
data_provider = image_util.ImageDataProvider("Model_zoo/train_aug/train/*.png",
data_suffix=".png",
mask_suffix='_mask.png',
n_class=2)
output_path = "TNBC2_CKPT"
net = unet.Unet(layers=4, features_root=6, channels=3, n_class=2)
trainer = unet.Trainer(net,
batch_size=8,
verification_batch_size=4,
optimizer="adam")
path = trainer.train(data_provider,
output_path,
keep_prob=0.8,
training_iters=32,
epochs=100,
display_step=2,
restore=True)
test_x, test_y = data_provider2(1)
prediction = net.predict(path, test_x)
error = unet.error_rate(prediction, util.crop_to_shape(test_y,
prediction.shape))
print(error)
img = util.combine_img_prediction(test_x, test_y, prediction)
util.save_image(img, "tnbc2.jpg")
import numpy as np
import os
# data_provider2 = image_util.ImageDataProvider("CHASE_NEW/test/*",data_suffix="R.jpg", mask_suffix='R_1stHO.png', n_class=2)
data_provider2 = image_util.ImageDataProvider("DRIVE700/test/*",data_suffix="_test.tif", mask_suffix='_manual1.png', n_class=2)
#data_provider2 = image_util.ImageDataProvider("DriveTest1/*",data_suffix="_test.tif", mask_suffix='_manual1.gif', n_class=2)
data_provider = image_util.ImageDataProvider("DRIVE700/train/*",data_suffix="_training.tif", mask_suffix='_manual1.png', n_class=2)
output_path="mode/drive100_0.92_700"
net = unet.Unet(layers=4, features_root=8, channels=3, n_class=2)
trainer = unet.Trainer(net,batch_size=4,verification_batch_size=4,optimizer="adam")
path = trainer.train(data_provider, output_path,keep_prob=0.92,training_iters=64, epochs=80,display_step=2,restore=False)
test_x, test_y = data_provider2(20)
prediction = net.predict(path, test_x)
prediction=util.crop_to_shape(prediction, (20,584,565,2))
AUC_ROC = metrics.roc_Auc(prediction, util.crop_to_shape(test_y, prediction.shape))
print("auc:",AUC_ROC)
acc=metrics.acc(prediction,util.crop_to_shape(test_y, prediction.shape))
print("acc:",acc)
precision=metrics.precision(prediction,util.crop_to_shape(test_y, prediction.shape))
print("ppv:",precision)
sen=metrics.sen(prediction,util.crop_to_shape(test_y, prediction.shape))
print("TPR:",sen)
TNR=metrics.TNR(prediction,util.crop_to_shape(test_y, prediction.shape))
print("tnr:",TNR)
f1=metrics.f1score2(prediction,util.crop_to_shape(test_y, prediction.shape))
print("f1:",f1)
img = util.combine_img_prediction(test_x, test_y, prediction)
util.save_image(img, "Drive_deep.jpg")
def train(self,
data_provider,
output_path,
training_iters=10,
epochs=100,
keep_prob=0.9,
block_size=7,
display_step=1,
restore=False,
write_graph=False,
prediction_path='prediction'):
"""
Lauches the training process
:param data_provider: callable returning training and verification data
:param output_path: path where to store checkpoints
:param training_iters: number of training mini batch iteration
:param epochs: number of epochs
:param dropout: dropout probability
:param display_step: number of steps till outputting stats
:param restore: Flag if previous model should be restored
:param write_graph: Flag if the computation graph should be written as protobuf file to the output path
:param prediction_path: path where to save predictions on each epoch
"""
save_path = os.path.join(output_path, "model.ckpt")
if epochs == 0:
return save_path
init = self._initialize(training_iters, output_path, restore,
prediction_path)
with tf.Session() as sess:
if write_graph:
tf.train.write_graph(sess.graph_def, output_path, "graph.pb",
False)
sess.run(init)
if restore:
ckpt = tf.train.get_checkpoint_state(output_path)
if ckpt and ckpt.model_checkpoint_path:
self.net.restore(sess, ckpt.model_checkpoint_path)
test_x, test_y = data_provider(self.verification_batch_size)
pred_shape = self.store_prediction(sess, test_x, test_y, "_init")
summary_writer = tf.summary.FileWriter(output_path,
graph=sess.graph)
logging.info("Start optimization")
avg_gradients = None
for epoch in range(epochs):
total_loss = 0
for step in range((epoch * training_iters),
((epoch + 1) * training_iters)):
batch_x, batch_y = data_provider(self.batch_size)
# Run optimization op (backprop)
_, loss, lr, gradients = sess.run(
(self.optimizer, self.net.cost,
self.learning_rate_node, self.net.gradients_node),
feed_dict={
self.net.x: batch_x,
self.net.y:
util.crop_to_shape(batch_y, pred_shape),
self.net.keep_prob: keep_prob,
self.net.block_size: block_size
})
if self.net.summaries and self.norm_grads:
avg_gradients = _update_avg_gradients(
avg_gradients, gradients, step)
norm_gradients = [
np.linalg.norm(gradient)
for gradient in avg_gradients
]
self.norm_gradients_node.assign(norm_gradients).eval()
if step % display_step == 0:
self.output_minibatch_stats(
sess, summary_writer, step, batch_x,
util.crop_to_shape(batch_y, pred_shape))
total_loss += loss
self.output_epoch_stats(epoch, total_loss, training_iters, lr)
self.store_prediction(sess, test_x, test_y, "epoch_%s" % epoch)
save_path = self.net.save(sess, save_path)
logging.info("Optimization Finished!")
return save_path
data_suffix=".tif",
mask_suffix='_binary.tif',
n_class=2)
#data_provider = image_util.ImageDataProvider("Tissue_images/*.tif",data_suffix=".tif", mask_suffix='_binary.tif', n_class=2)
data_provider = image_util.ImageDataProvider("tissue_aug/train/*.tif",
data_suffix=".tif",
mask_suffix='_binary.tif',
n_class=2)
output_path = "out_put_skip_aug3_160"
#setup & training
net = unet2.Unet(layers=4, features_root=8, channels=3, n_class=2)
trainer = unet2.Trainer(net,
batch_size=4,
verification_batch_size=4,
optimizer="adam")
path = trainer.train(data_provider,
output_path,
keep_prob=0.8,
training_iters=124,
epochs=1,
display_step=2,
restore=True)
test_x, test_y = data_provider2(14)
prediction = net.predict(path, test_x)
print(test_y)
error = unet2.error_rate(prediction,
util.crop_to_shape(test_y, prediction.shape))
print("test error:", error)
img = util.combine_img_prediction(test_x, test_y, prediction)
util.save_image(img, "test_aug_5.jpg")
shuffle_data=False)
net = unet.Unet(layers=4, features_root=8, channels=3, n_class=2)
test_x, test_y = DT(6)
model = "model/model_f8_0.88_flip_160/model.ckpt"
prediction = net.predict(model, test_x)
#
# pred=tf.cast(prediction, tf.float64)
# loss=stats_utils.get_dice(util.crop_to_shape(test_y, prediction.shape),prediction)
# init = tf.global_variables_initializer()
# with tf.Session() as sess:
# # Initialize variables
# sess.run(init)
# print("loss",sess.run(loss))
print(
"dice2:",
stats_utils.get_dice_2(util.crop_to_shape(test_y, prediction.shape),
prediction))
print(
"dice1",
stats_utils.get_dice_1(util.crop_to_shape(test_y, prediction.shape),
prediction))
print("f1:",
unet.f1score2(prediction, util.crop_to_shape(test_y, prediction.shape)))
print(
"aji:",
stats_utils.get_aji(util.crop_to_shape(test_y, prediction.shape),
prediction))
print(
"aji+:",
stats_utils.get_fast_aji(util.crop_to_shape(test_y, prediction.shape),
prediction))
from tf_SDUnet import unet, util, image_util,stats_utils,layers
import tensorflow as tf
import numpy as np
Single = image_util.ImageDataProvider("test2/single/*.tif",data_suffix=".tif", mask_suffix='_binary.tif', n_class=2,shuffle_data=False)
net = unet.Unet(layers=4, features_root=8, channels=3, n_class=2)
test_x, test_y =Single(1)
model="model/model_f8_0.88_flip_160/model.ckpt"
prediction = net.predict(model, test_x)
print("dice2:",stats_utils.get_dice_2(util.crop_to_shape(test_y, prediction.shape),prediction))
print("dice1",stats_utils.get_dice_1(util.crop_to_shape(test_y, prediction.shape),prediction))
print("f1:",unet.f1score2(prediction, util.crop_to_shape(test_y, prediction.shape)))
print("aji:",stats_utils.get_aji(util.crop_to_shape(test_y, prediction.shape),prediction))
print("aji+:",stats_utils.get_fast_aji(util.crop_to_shape(test_y, prediction.shape),prediction))
error=unet.error_rate(prediction, util.crop_to_shape(test_y, prediction.shape))
print("error:",error)
img = util.combine_img_prediction(test_x, test_y, prediction)
util.save_image(img, "result/5698.jpg")
cost="dice_coefficient")
trainer = unet.Trainer(net,
batch_size=4,
verification_batch_size=4,
optimizer="adam")
path = trainer.train(data_provider,
output_path,
keep_prob=0.88,
block_size=7,
training_iters=64,
epochs=100,
display_step=2,
restore=False)
test_x, test_y = DT(6)
prediction = net.predict(path, test_x)
error = unet.error_rate(prediction, util.crop_to_shape(test_y,
prediction.shape))
print("DT error:", error)
f1 = unet.f1score2(prediction, util.crop_to_shape(test_y, prediction.shape))
print("DTf1:", f1)
img = util.combine_img_prediction(test_x, test_y, prediction)
util.save_image(img, "DTtest.jpg")
test_x, test_y = ST(8)
prediction = net.predict(path, test_x)
error = unet.error_rate(prediction, util.crop_to_shape(test_y,
prediction.shape))
print("ST error:", error)
f1 = unet.f1score2(prediction, util.crop_to_shape(test_y, prediction.shape))
print("STf1:", f1)
img = util.combine_img_prediction(test_x, test_y, prediction)
util.save_image(img, "STtest.jpg")