85 lines
4.1 KiB
Python
85 lines
4.1 KiB
Python
import tensorflow as tf
|
|
import numpy as np
|
|
|
|
def convolution(couche_prec, taille_noyau, nbr_noyau):
|
|
w=tf.Variable(tf.random.truncated_normal(shape=(taille_noyau, taille_noyau, int(couche_prec.get_shape()[-1]), nbr_noyau)))
|
|
b=np.zeros(nbr_noyau)
|
|
result=tf.nn.conv2d(couche_prec, w, strides=[1, 1, 1, 1], padding='SAME')+b
|
|
return result
|
|
|
|
def fc(couche_prec, nbr_neurone):
|
|
w=tf.Variable(tf.random.truncated_normal(shape=(int(couche_prec.get_shape()[-1]), nbr_neurone), dtype=tf.float32))
|
|
b=tf.Variable(np.zeros(shape=(nbr_neurone)), dtype=tf.float32)
|
|
result=tf.matmul(couche_prec, w)+b
|
|
return result
|
|
|
|
def vggnet(learning_rate=0.01, momentum=0.99):
|
|
ph_images=tf.placeholder(shape=(None, 28, 28, 1), dtype=tf.float32, name='images')
|
|
ph_labels=tf.placeholder(shape=(None, 10), dtype=tf.float32)
|
|
ph_is_training=tf.placeholder_with_default(False, (), name='is_training')
|
|
|
|
result=convolution(ph_images, 3, 64)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=convolution(result, 3, 64)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=tf.nn.max_pool(result, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
|
|
|
|
result=convolution(result, 3, 128)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=convolution(result, 3, 128)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=tf.nn.max_pool(result, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
|
|
|
|
result=convolution(result, 3, 256)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=convolution(result, 3, 256)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=convolution(result, 3, 256)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=tf.nn.max_pool(result, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
|
|
|
|
result=convolution(result, 3, 512)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=convolution(result, 3, 512)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=convolution(result, 3, 512)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=tf.nn.max_pool(result, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
|
|
|
|
result=convolution(result, 3, 512)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=convolution(result, 3, 512)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=convolution(result, 3, 512)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=tf.nn.max_pool(result, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
|
|
result=tf.contrib.layers.flatten(result)
|
|
|
|
result=fc(result, 512)
|
|
result=tf.layers.batch_normalization(result, training=ph_is_training, momentum=momentum)
|
|
result=tf.nn.relu(result)
|
|
result=fc(result, 10)
|
|
socs=tf.nn.softmax(result, name="sortie")
|
|
|
|
loss=tf.nn.softmax_cross_entropy_with_logits_v2(labels=ph_labels, logits=result)
|
|
extra_update_ops=tf.get_collection(tf.GraphKeys.UPDATE_OPS)
|
|
with tf.control_dependencies(extra_update_ops):
|
|
train=tf.train.AdamOptimizer(learning_rate).minimize(loss)
|
|
accuracy=tf.reduce_mean(tf.cast(tf.equal(tf.argmax(socs, 1), tf.argmax(ph_labels, 1)), tf.float32))
|
|
|
|
return ph_images, ph_labels, ph_is_training, socs, train, accuracy, tf.train.Saver()
|
|
|