def extract_features(images, output_size, use_batch_norm, dropout_keep_prob):
"""
Based on the architecture described in 'Matching Networks for One-Shot Learning'
http://arxiv.org/abs/1606.04080.pdf.
:param images: batch of images.
:param output_size: dimensionality of the output features.
:param use_batch_norm: whether to use batch normalization or not.
:param dropout_keep_prob: keep probability parameter for dropout.
:return: features.
"""
# 4X conv2d + pool blocks
h = conv2d_pool_block(images, use_batch_norm, dropout_keep_prob,
'fe_block_1')
h = conv2d_pool_block(h, use_batch_norm, dropout_keep_prob, 'fe_block_2')
h = conv2d_pool_block(h, use_batch_norm, dropout_keep_prob, 'fe_block_3')
h = conv2d_pool_block(h, use_batch_norm, dropout_keep_prob, 'fe_block_4')
# flatten output
h = tf.contrib.layers.flatten(h)
# dense layer
h = dense_block(h, output_size, use_batch_norm, dropout_keep_prob,
'fe_dense')
return h
def extract_features_mini_imagenet(images, output_size, use_batch_norm, rate):
"""
Based on the architecture described in 'Matching Networks for One-Shot Learning'
http://arxiv.org/abs/1606.04080.pdf.
:param images: batch of images.
:param output_size: dimensionality of the output features.
:param use_batch_norm: whether to use batch normalization or not.
:param rate: drop probability parameter for dropout.
:return: features.
"""
# 5X conv2d + pool blocks
h = conv2d_pool_block(images, use_batch_norm, rate, 'valid', 'fe_block_1')
h = conv2d_pool_block(h, use_batch_norm, rate, 'valid', 'fe_block_2')
h = conv2d_pool_block(h, use_batch_norm, rate, 'valid', 'fe_block_3')
h = conv2d_pool_block(h, use_batch_norm, rate, 'valid', 'fe_block_4')
h = conv2d_pool_block(h, use_batch_norm, rate, 'valid', 'fe_block_5')
# flatten output
h = tf.compat.v1.layers.flatten(h)
return h
