Skip to content

GitLab

otbtf
otbtf
Commit 8cc0e197 authored by Cresson Remi's avatar Cresson Remi
Browse files
Options

WIP: migrade python codes from TF1.X --> TF2.X

parent 30634814
Hide whitespace changes
Inline Side-by-side
Showing with 479 additions and 507 deletions
python/ckpt2savedmodel.py
View file @ 8cc0e197
# -*- coding: utf-8 -*-
#==========================================================================
# ==========================================================================
#
# Copyright 2018-2019 Remi Cresson (IRSTEA)
# Copyright 2020 Remi Cresson (INRAE)
......@@ -16,28 +16,24 @@
# See the License for the specific language governing permissions and
# limitations under the License.
#
#==========================================================================*/
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
# ==========================================================================*/
import argparse
from tricks import CheckpointToSavedModel
from tricks import ckpt_to_savedmodel
# Parser
parser = argparse.ArgumentParser()
parser.add_argument("--ckpt", help="Checkpoint file (without the \".meta\" extension)", required=True)
parser.add_argument("--inputs", help="Inputs names (e.g. [\"x_cnn_1:0\", \"x_cnn_2:0\"])", required=True, nargs='+')
parser.add_argument("--outputs", help="Outputs names (e.g. [\"prediction:0\", \"features:0\"])", required=True, nargs='+')
parser.add_argument("--model", help="Output directory for SavedModel", required=True)
parser.add_argument("--ckpt", help="Checkpoint file (without the \".meta\" extension)", required=True)
parser.add_argument("--inputs", help="Inputs names (e.g. [\"x_cnn_1:0\", \"x_cnn_2:0\"])", required=True, nargs='+')
parser.add_argument("--outputs", help="Outputs names (e.g. [\"prediction:0\", \"features:0\"])", required=True,
nargs='+')
parser.add_argument("--model", help="Output directory for SavedModel", required=True)
parser.add_argument('--clear_devices', dest='clear_devices', action='store_true')
parser.set_defaults(clear_devices=False)
params = parser.parse_args()
if __name__ == "__main__":
CheckpointToSavedModel(ckpt_path=params.ckpt,
inputs=params.inputs,
outputs=params.outputs,
savedmodel_path=params.model,
clear_devices=params.clear_devices)
quit()
ckpt_to_savedmodel(ckpt_path=params.ckpt,
inputs=params.inputs,
outputs=params.outputs,
savedmodel_path=params.model,
clear_devices=params.clear_devices)
python/create_savedmodel_ienco-m3_patchbased.py
View file @ 8cc0e197
# -*- coding: utf-8 -*-
#==========================================================================
# ==========================================================================
#
# Copyright Remi Cresson, Dino Ienco (IRSTEA)
#
......@@ -15,208 +15,196 @@
# See the License for the specific language governing permissions and
# limitations under the License.
#
#==========================================================================*/
# ==========================================================================*/
# Reference:
#
# Benedetti, P., Ienco, D., Gaetano, R., Ose, K., Pensa, R. G., & Dupuy, S. (2018)
# M3Fusion: A Deep Learning Architecture for Multiscale Multimodal Multitemporal
# Satellite Data Fusion. IEEE Journal of Selected Topics in Applied Earth
# M3Fusion: A Deep Learning Architecture for Multiscale Multimodal Multitemporal
# Satellite Data Fusion. IEEE Journal of Selected Topics in Applied Earth
# Observations and Remote Sensing, 11(12), 4939-4949.
from tricks import *
from tensorflow.contrib import rnn
from tricks import create_savedmodel
import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()
from tf.contrib import rnn
import argparse
parser = argparse.ArgumentParser()
parser.add_argument("--nunits", type=int, default=1024, help="number of units")
parser.add_argument("--n_levels_lstm", type=int, default=1, help="number of lstm levels")
parser.add_argument("--hm_epochs", type=int, default=400, help="hm epochs")
parser.add_argument("--n_timestamps", type=int, default=37, help="number of images in timeseries")
parser.add_argument("--n_dims", type=int, default=16, help="number of channels in timeseries images")
parser.add_argument("--patch_window", type=int, default=25, help="patch size for the high-res image")
parser.add_argument("--n_channels", type=int, default=4, help="number of channels in the high-res image")
parser.add_argument("--nclasses", type=int, default=8, help="number of classes")
parser.add_argument("--nunits", type=int, default=1024, help="number of units")
parser.add_argument("--n_levels_lstm", type=int, default=1, help="number of lstm levels")
parser.add_argument("--hm_epochs", type=int, default=400, help="hm epochs")
parser.add_argument("--n_timestamps", type=int, default=37, help="number of images in timeseries")
parser.add_argument("--n_dims", type=int, default=16, help="number of channels in timeseries images")
parser.add_argument("--patch_window", type=int, default=25, help="patch size for the high-res image")
parser.add_argument("--n_channels", type=int, default=4, help="number of channels in the high-res image")
parser.add_argument("--nclasses", type=int, default=8, help="number of classes")
parser.add_argument("--outdir", help="Output directory for SavedModel", required=True)
params = parser.parse_args()
def RnnAttention(x, nunits, nlayer, n_dims, n_timetamps, is_training_ph):
N = tf.shape(x)[0] # size of batch
x = tf.reshape(x, [N, n_dims, n_timetamps])
# at this point x must be 1 tensor of shape [N, n_dims, n_timestamps]
# (before unstack) x is 1 tensor of shape [N, n_dims, n_timestamps]
x = tf.unstack(x, n_timetamps, axis=2)
# (after unstack) x is a list of "n_timestamps" tensors of shape: [N, n_dims]
# NETWORK DEF
# MORE THEN ONE LAYER: list of LSTMcell,nunits hidden units each, for each layer
if nlayer > 1:
cells = []
for _ in range(nlayer):
cell = rnn.GRUCell(nunits)
cells.append(cell)
cell = tf.compat.v1.contrib.rnn.MultiRNNCell(cells)
# SIGNLE LAYER: single GRUCell, nunits hidden units each
else:
cell = rnn.GRUCell(nunits)
outputs, _ = rnn.static_rnn(cell, x, dtype="float32")
# At this point, outputs is a list of "n_timestamps" tensors [N, B, C]
outputs = tf.stack(outputs, axis=1)
# At this point, outputs is a tensor of size [N, n_timestamps, B, C]
# Trainable parameters
attention_size = nunits # int(nunits / 2)
W_omega = tf.Variable(tf.random_normal([nunits, attention_size], stddev=0.1))
b_omega = tf.Variable(tf.random_normal([attention_size], stddev=0.1))
u_omega = tf.Variable(tf.random_normal([attention_size], stddev=0.1))
# Applying fully connected layer with non-linear activation to each of the B*T timestamps;
# the shape of `v` is (B,T,D)*(D,A)=(B,T,A), where A=attention_size
v = tf.tanh(tf.tensordot(outputs, W_omega, axes=1) + b_omega)
# For each of the timestamps its vector of size A from `v` is reduced with `u` vector
vu = tf.tensordot(v, u_omega, axes=1) # (B,T) shape
alphas = tf.nn.softmax(vu) # (B,T) shape also
output = tf.reduce_sum(outputs * tf.expand_dims(alphas, -1), 1)
output = tf.reshape(output, [-1, nunits])
return output
N = tf.shape(x)[0] # size of batch
x = tf.reshape(x, [N, n_dims, n_timetamps])
# at this point x must be 1 tensor of shape [N, n_dims, n_timestamps]
# (before unstack) x is 1 tensor of shape [N, n_dims, n_timestamps]
x = tf.unstack(x, n_timetamps, axis=2)
# (after unstack) x is a list of "n_timestamps" tensors of shape: [N, n_dims]
#NETWORK DEF
#MORE THEN ONE LAYER: list of LSTMcell,nunits hidden units each, for each layer
if nlayer>1:
cells=[]
for _ in range(nlayer):
cell = rnn.GRUCell(nunits)
cells.append(cell)
cell = tf.contrib.rnn.MultiRNNCell(cells)
#SIGNLE LAYER: single GRUCell, nunits hidden units each
else:
cell = rnn.GRUCell(nunits)
outputs,_=rnn.static_rnn(cell, x, dtype="float32")
# At this point, outputs is a list of "n_timestamps" tensors [N, B, C]
outputs = tf.stack(outputs, axis=1)
# At this point, outputs is a tensor of size [N, n_timestamps, B, C]
# Trainable parameters
attention_size = nunits #int(nunits / 2)
W_omega = tf.Variable(tf.random_normal([nunits, attention_size], stddev=0.1))
b_omega = tf.Variable(tf.random_normal([attention_size], stddev=0.1))
u_omega = tf.Variable(tf.random_normal([attention_size], stddev=0.1))
# Applying fully connected layer with non-linear activation to each of the B*T timestamps;
# the shape of `v` is (B,T,D)*(D,A)=(B,T,A), where A=attention_size
v = tf.tanh(tf.tensordot(outputs, W_omega, axes=1) + b_omega)
# For each of the timestamps its vector of size A from `v` is reduced with `u` vector
vu = tf.tensordot(v, u_omega, axes=1) # (B,T) shape
alphas = tf.nn.softmax(vu) # (B,T) shape also
output = tf.reduce_sum(outputs * tf.expand_dims(alphas, -1), 1)
output = tf.reshape(output, [-1, nunits])
return output
def CNN(x, nunits):
#nunits = 512
conv1 = tf.layers.conv2d(
# nunits = 512
conv1 = tf.compat.v1.layers.conv2d(
inputs=x,
filters=nunits/2, #256
filters=nunits / 2, # 256
kernel_size=[7, 7],
padding="valid",
activation=tf.nn.relu)
conv1 = tf.layers.batch_normalization(conv1)
print_tensor_info("conv1", conv1)
pool1 = tf.layers.max_pooling2d(inputs=conv1, pool_size=[2, 2], strides=2)
print_tensor_info("pool1", pool1)
conv2 = tf.layers.conv2d(
conv1 = tf.compat.v1.layers.batch_normalization(conv1)
pool1 = tf.compat.v1.layers.max_pooling2d(inputs=conv1, pool_size=[2, 2], strides=2)
conv2 = tf.compat.v1.layers.conv2d(
inputs=pool1,
filters=nunits,
kernel_size=[3, 3],
padding="valid",
activation=tf.nn.relu)
conv2 = tf.layers.batch_normalization(conv2)
print_tensor_info("conv2", conv2)
conv3 = tf.layers.conv2d(
conv2 = tf.compat.v1.layers.batch_normalization(conv2)
conv3 = tf.compat.v1.layers.conv2d(
inputs=conv2,
filters=nunits,
kernel_size=[3, 3],
padding="same",
activation=tf.nn.relu)
conv3 = tf.layers.batch_normalization(conv3)
print_tensor_info("conv3", conv3)
conv3 = tf.concat([conv2,conv3],3)
print_tensor_info("conv3 (final)", conv3)
conv4 = tf.layers.conv2d(
conv3 = tf.compat.v1.layers.batch_normalization(conv3)
conv3 = tf.compat.v1.concat([conv2, conv3], 3)
conv4 = tf.compat.v1.layers.conv2d(
inputs=conv3,
filters=nunits,
kernel_size=[1, 1],
padding="valid",
activation=tf.nn.relu)
conv4 = tf.layers.batch_normalization(conv4)
print_tensor_info("conv4", conv4)
cnn = tf.reduce_mean(conv4, [1,2])
print_tensor_info("cnn", cnn)
tensor_shape = cnn.get_shape()
return cnn, tensor_shape[1].value
def getPrediction(x_rnn, x_cnn, nunits, nlayer, nclasses, dropout, is_training, n_dims, n_timetamps):
features_learnt = None
vec_rnn = RnnAttention(x_rnn, nunits, nlayer, n_dims, n_timetamps, is_training_ph)
vec_cnn, cnn_dim = CNN(x_cnn, 512)
features_learnt=tf.concat([vec_rnn,vec_cnn],axis=1, name="features")
first_dim = cnn_dim + nunits
#Classifier1 #RNN Branch
outb1 = tf.Variable(tf.truncated_normal([nclasses]),name='B1')
outw1 = tf.Variable(tf.truncated_normal([nunits,nclasses]),name='W1')
pred_c1 = tf.matmul(vec_rnn,outw1)+outb1
#Classifier2 #CNN Branch
outb2 = tf.Variable(tf.truncated_normal([nclasses]),name='B2')
outw2 = tf.Variable(tf.truncated_normal([cnn_dim,nclasses]),name='W2')
pred_c2 = tf.matmul(vec_cnn,outw2)+outb2
#ClassifierFull
outb = tf.Variable(tf.truncated_normal([nclasses]),name='B')
outw = tf.Variable(tf.truncated_normal([first_dim,nclasses]),name='W')
pred_full = tf.matmul(features_learnt,outw)+outb
return pred_c1, pred_c2, pred_full, features_learnt
###############################################################################
""" Main """
conv4 = tf.compat.v1.layers.batch_normalization(conv4)
cnn = tf.reduce_mean(conv4, [1, 2])
tensor_shape = cnn.get_shape()
return cnn, tensor_shape[1].value
def get_prediction(x_rnn, x_cnn, nunits, nlayer, nclasses, n_dims, n_timetamps):
vec_rnn = RnnAttention(x_rnn, nunits, nlayer, n_dims, n_timetamps, is_training_ph)
vec_cnn, cnn_dim = CNN(x_cnn, 512)
features_learnt = tf.concat([vec_rnn, vec_cnn], axis=1, name="features")
first_dim = cnn_dim + nunits
# Classifier1 #RNN Branch
outb1 = tf.Variable(tf.truncated_normal([nclasses]), name='B1')
outw1 = tf.Variable(tf.truncated_normal([nunits, nclasses]), name='W1')
pred_c1 = tf.matmul(vec_rnn, outw1) + outb1
# Classifier2 #CNN Branch
outb2 = tf.Variable(tf.truncated_normal([nclasses]), name='B2')
outw2 = tf.Variable(tf.truncated_normal([cnn_dim, nclasses]), name='W2')
pred_c2 = tf.matmul(vec_cnn, outw2) + outb2
# ClassifierFull
outb = tf.Variable(tf.truncated_normal([nclasses]), name='B')
outw = tf.Variable(tf.truncated_normal([first_dim, nclasses]), name='W')
pred_full = tf.matmul(features_learnt, outw) + outb
return pred_c1, pred_c2, pred_full, features_learnt
# Create the TensorFlow graph
with tf.Graph().as_default():
x_rnn = tf.placeholder(tf.float32, [None, 1, 1, params.n_dims*params.n_timestamps], name="x_rnn")
x_cnn = tf.placeholder(tf.float32, [None, params.patch_window, params.patch_window, params.n_channels], name="x_cnn")
y = tf.placeholder(tf.int32, [None, 1, 1, 1], name="y")
learning_rate = tf.placeholder_with_default(tf.constant(0.0002, dtype=tf.float32, shape=[]), shape=[], name="learning_rate")
is_training_ph = tf.placeholder_with_default(tf.constant(False, dtype=tf.bool, shape=[]), shape=[], name="is_training")
dropout = tf.placeholder_with_default(tf.constant(0.5, dtype=tf.float32, shape=[]), shape=[], name="drop_rate")
pred_c1, pred_c2, pred_full, features_learnt = getPrediction(x_rnn,
x_cnn,
params.nunits,
params.n_levels_lstm,
params.nclasses,
dropout,
is_training_ph,
params.n_dims,
params.n_timestamps)
testPrediction = tf.argmax(pred_full, 1, name="prediction")
loss_full = tf.losses.sparse_softmax_cross_entropy(labels=tf.reshape(y, [-1, 1]),
logits=tf.reshape(pred_full, [-1, params.nclasses]))
loss_c1 = tf.losses.sparse_softmax_cross_entropy(labels=tf.reshape(y, [-1, 1]),
logits=tf.reshape(pred_c1, [-1, params.nclasses]))
loss_c2 = tf.losses.sparse_softmax_cross_entropy(labels=tf.reshape(y, [-1, 1]),
logits=tf.reshape(pred_c2, [-1, params.nclasses]))
cost = loss_full + (0.3 * loss_c1) + (0.3 * loss_c2)
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate, name="optimizer").minimize(cost)
correct = tf.equal(tf.argmax(pred_full,1),tf.argmax(y,1))
accuracy = tf.reduce_mean(tf.cast(correct,tf.float64))
# Initializer, saver, session
init = tf.global_variables_initializer()
saver = tf.train.Saver( max_to_keep=20 )
sess = tf.Session()
sess.run(init)
CreateSavedModel(sess, ["x_cnn:0","x_rnn:0","y:0"], ["prediction:0"], params.outdir)
with tf.compat.v1.Graph().as_default():
x_rnn = tf.compat.v1.placeholder(tf.float32, [None, 1, 1, params.n_dims * params.n_timestamps], name="x_rnn")
x_cnn = tf.compat.v1.placeholder(tf.float32, [None, params.patch_window, params.patch_window, params.n_channels],
name="x_cnn")
y = tf.compat.v1.placeholder(tf.int32, [None, 1, 1, 1], name="y")
learning_rate = tf.compat.v1.placeholder_with_default(tf.constant(0.0002, dtype=tf.float32, shape=[]), shape=[],
name="learning_rate")
is_training_ph = tf.compat.v1.placeholder_with_default(tf.constant(False, dtype=tf.bool, shape=[]), shape=[],
name="is_training")
dropout = tf.compat.v1.placeholder_with_default(tf.constant(0.5, dtype=tf.float32, shape=[]), shape=[],
name="drop_rate")
pred_c1, pred_c2, pred_full, features_learnt = get_prediction(x_rnn,
x_cnn,
params.nunits,
params.n_levels_lstm,
params.nclasses,
params.n_dims,
params.n_timestamps)
testPrediction = tf.argmax(pred_full, 1, name="prediction")
loss_full = tf.compat.v1.losses.sparse_softmax_cross_entropy(labels=tf.reshape(y, [-1, 1]),
logits=tf.reshape(pred_full, [-1, params.nclasses]))
loss_c1 = tf.compat.v1.losses.sparse_softmax_cross_entropy(labels=tf.reshape(y, [-1, 1]),
logits=tf.reshape(pred_c1, [-1, params.nclasses]))
loss_c2 = tf.compat.v1.losses.sparse_softmax_cross_entropy(labels=tf.reshape(y, [-1, 1]),
logits=tf.reshape(pred_c2, [-1, params.nclasses]))
cost = loss_full + (0.3 * loss_c1) + (0.3 * loss_c2)
optimizer = tf.compat.v1.train.AdamOptimizer(learning_rate=learning_rate, name="optimizer").minimize(cost)
correct = tf.equal(tf.argmax(pred_full, 1), tf.argmax(y, 1))
accuracy = tf.reduce_mean(tf.cast(correct, tf.float64))
# Initializer, saver, session
init = tf.compat.v1.global_variables_initializer()
saver = tf.compat.v1.train.Saver(max_to_keep=20)
sess = tf.compat.v1.Session()
sess.run(init)
create_savedmodel(sess, ["x_cnn:0", "x_rnn:0", "y:0"], ["prediction:0"], params.outdir)
python/create_savedmodel_maggiori17_fullyconv.py
View file @ 8cc0e197
# -*- coding: utf-8 -*-
#==========================================================================
#
# Copyright Remi Cresson (IRSTEA)
# Copyright 2018-2019 Remi Cresson (IRSTEA)
# Copyright 2020 Remi Cresson (INRAE)
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
......@@ -19,12 +20,14 @@
# Reference:
#
# Maggiori, E., Tarabalka, Y., Charpiat, G., & Alliez, P. (2016).
# Maggiori, E., Tarabalka, Y., Charpiat, G., & Alliez, P. (2016).
# "Convolutional neural networks for large-scale remote-sensing image classification."
# IEEE Transactions on Geoscience and Remote Sensing, 55(2), 645-657.
from tricks import *
import argparse
from tricks import create_savedmodel
import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()
parser = argparse.ArgumentParser()
parser.add_argument("--outdir", help="Output directory for SavedModel", required=True)
......@@ -32,93 +35,69 @@ parser.add_argument("--n_channels", type=int, default=4, help="number of channel
params = parser.parse_args()
# Build the graph
with tf.Graph().as_default():
# Size of patches
patch_size_xs = 80
patch_size_label = 16
# placeholder for images and labels
istraining_placeholder = tf.placeholder_with_default(tf.constant(False , dtype=tf.bool, shape=[]), shape=[], name="is_training")
learning_rate = tf.placeholder_with_default(tf.constant(0.0002, dtype=tf.float32, shape=[]), shape=[], name="learning_rate")
xs_placeholder = tf.placeholder(tf.float32, shape=(None, patch_size_xs, patch_size_xs, params.n_channels), name="x")
labels_placeholder = tf.placeholder(tf.int32, shape=(None, patch_size_label, patch_size_label, 1), name="y")
# Convolutional Layer #1
conv1 = tf.layers.conv2d(
inputs=xs_placeholder,
filters=64,
kernel_size=[12, 12],
padding="valid",
activation=tf.nn.crelu)
# Normalization of output of layer 1
norm1 = tf.layers.batch_normalization(conv1)
# pooling layer #1
pool1 = tf.layers.max_pooling2d(inputs=norm1, pool_size=[4, 4], strides=4)
# Convolutional Layer #2
conv2 = tf.layers.conv2d(
inputs=pool1,
filters=112,
kernel_size=[4, 4],
padding="valid",
activation=tf.nn.crelu)
# Normalization of output of layer 2
norm2 = tf.layers.batch_normalization(conv2)
# Convolutional Layer #3
conv3 = tf.layers.conv2d(
inputs=norm2,
filters=80,
kernel_size=[3, 3],
padding="valid",
activation=tf.nn.crelu)
# Normalization of output of layer 3
norm3 = tf.layers.batch_normalization(conv3)
# Convolutional Layer #4
conv4 = tf.layers.conv2d(
inputs=norm3,
filters=1,
kernel_size=[8, 8],
padding="valid",
activation=tf.nn.crelu)
print_tensor_info("conv4",conv4)
# Deconv = conv on the padded/strided input, that is an (5+1)*4
deconv1 = tf.layers.conv2d_transpose(
inputs=conv4,
filters=1,
strides=(4,4),
kernel_size=[8, 8],
padding="valid",
activation=tf.nn.sigmoid)
print_tensor_info("deconv1",deconv1)
numbatch = tf.shape(deconv1)[0]
szx = tf.shape(deconv1)[1]
szy = tf.shape(deconv1)[2]
estimated = tf.slice(deconv1, [0, 4, 4, 0], [numbatch, szx - 8, szy - 8, 1], "estimated")
print_tensor_info("estimated", estimated)
# Loss
estimated_resized = tf.reshape(estimated, [-1, patch_size_label*patch_size_label])
labels_resized = tf.reshape(labels_placeholder, [-1, patch_size_label*patch_size_label])
labels_resized = tf.cast(labels_resized, tf.float32)
loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(labels=labels_resized, logits=estimated_resized))
# Optimizer
train_op = tf.train.AdamOptimizer(learning_rate, name="optimizer").minimize(loss)
# Initializer, saver, session
init = tf.global_variables_initializer()
saver = tf.train.Saver( max_to_keep=20 )
sess = tf.Session()
sess.run(init)
# Let's export a SavedModel
CreateSavedModel(sess, ["x:0", "y:0", "is_training:0"], ["estimated:0"], params.outdir)
\ No newline at end of file
with tf.compat.v1.Graph().as_default():
# Size of patches
patch_size_xs = 80
patch_size_label = 16
# placeholder for images and labels
lr = tf.compat.v1.placeholder_with_default(tf.constant(0.0002, dtype=tf.float32, shape=[]), shape=[],
name="learning_rate")
x = tf.compat.v1.placeholder(tf.float32, shape=(None, patch_size_xs, patch_size_xs, params.n_channels), name="x")
y = tf.compat.v1.placeholder(tf.int32, shape=(None, patch_size_label, patch_size_label, 1), name="y")
# Convolutional Layer #1
conv1 = tf.compat.v1.layers.conv2d(inputs=x, filters=64, kernel_size=[12, 12], padding="valid",
activation=tf.nn.crelu)
# Normalization of output of layer 1
norm1 = tf.compat.v1.layers.batch_normalization(conv1)
# pooling layer #1
pool1 = tf.compat.v1.layers.max_pooling2d(inputs=norm1, pool_size=[4, 4], strides=4)
# Convolutional Layer #2
conv2 = tf.compat.v1.layers.conv2d(inputs=pool1, filters=112, kernel_size=[4, 4], padding="valid",
activation=tf.nn.crelu)
# Normalization of output of layer 2
norm2 = tf.compat.v1.layers.batch_normalization(conv2)
# Convolutional Layer #3
conv3 = tf.compat.v1.layers.conv2d(inputs=norm2, filters=80, kernel_size=[3, 3], padding="valid",
activation=tf.nn.crelu)
# Normalization of output of layer 3
norm3 = tf.compat.v1.layers.batch_normalization(conv3)
# Convolutional Layer #4
conv4 = tf.compat.v1.layers.conv2d(inputs=norm3, filters=1, kernel_size=[8, 8], padding="valid",
activation=tf.nn.crelu)
# Deconv = conv on the padded/strided input, that is an (5+1)*4
deconv1 = tf.compat.v1.layers.conv2d_transpose(inputs=conv4, filters=1, strides=(4,4), kernel_size=[8, 8],
padding="valid", activation=tf.nn.sigmoid)
n = tf.shape(deconv1)[0]
szx = tf.shape(deconv1)[1]
szy = tf.shape(deconv1)[2]
estimated = tf.slice(deconv1, [0, 4, 4, 0], [n, szx - 8, szy - 8, 1], "estimated")
# Loss
estimated_resized = tf.reshape(estimated, [-1, patch_size_label*patch_size_label])
labels_resized = tf.reshape(y, [-1, patch_size_label*patch_size_label])
labels_resized = tf.cast(labels_resized, tf.float32)
loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(labels=labels_resized, logits=estimated_resized))