Function block_inception_c

tests/utils/custom_layers/inception_blocks.py:212–279

Builds Inception-C block for Inception v4 network.

(inputs, scope=None, is_train=False)

def block_inception_c(inputs, scope=None, is_train=False):
    """Builds Inception-C block for Inception v4 network."""
    # By default use stride=1 and SAME padding

    with tf.variable_scope(scope, 'BlockInceptionC', [inputs]):
        with tf.variable_scope('Branch_0'):
            branch_0, _ = conv_module(
                inputs, n_out_channel=256, filter_size=(1, 1), strides=(1, 1), padding='SAME', batch_norm_init=None,
                is_train=is_train, use_batchnorm=True, activation_fn='ReLU', name='Conv2d_0a_1x1'
            )

        with tf.variable_scope('Branch_1'):
            branch_1, _ = conv_module(
                inputs, n_out_channel=384, filter_size=(1, 1), strides=(1, 1), padding='SAME', batch_norm_init=None,
                is_train=is_train, use_batchnorm=True, activation_fn='ReLU', name='Conv2d_0a_1x1'
            )

            branch_1a, _ = conv_module(
                branch_1, n_out_channel=256, filter_size=(1, 3), strides=(1, 1), padding='SAME', batch_norm_init=None,
                is_train=is_train, use_batchnorm=True, activation_fn='ReLU', name='Conv2d_0b_1x3'
            )

            branch_1b, _ = conv_module(
                branch_1, n_out_channel=256, filter_size=(3, 1), strides=(1, 1), padding='SAME', batch_norm_init=None,
                is_train=is_train, use_batchnorm=True, activation_fn='ReLU', name='Conv2d_0c_3x1'
            )

            branch_1 = tl.layers.ConcatLayer([branch_1a, branch_1b], concat_dim=3, name='concat_layer')

        with tf.variable_scope('Branch_2'):
            branch_2, _ = conv_module(
                inputs, n_out_channel=384, filter_size=(1, 1), strides=(1, 1), padding='SAME', batch_norm_init=None,
                is_train=is_train, use_batchnorm=True, activation_fn='ReLU', name='Conv2d_0a_1x1'
            )

            branch_2, _ = conv_module(
                branch_2, n_out_channel=448, filter_size=(3, 1), strides=(1, 1), padding='SAME', batch_norm_init=None,
                is_train=is_train, use_batchnorm=True, activation_fn='ReLU', name='Conv2d_0b_3x1'
            )

            branch_2, _ = conv_module(
                branch_2, n_out_channel=512, filter_size=(1, 3), strides=(1, 1), padding='SAME', batch_norm_init=None,
                is_train=is_train, use_batchnorm=True, activation_fn='ReLU', name='Conv2d_0c_1x3'
            )

            branch_2a, _ = conv_module(
                branch_2, n_out_channel=256, filter_size=(1, 3), strides=(1, 1), padding='SAME', batch_norm_init=None,
                is_train=is_train, use_batchnorm=True, activation_fn='ReLU', name='Conv2d_0d_1x3'
            )

            branch_2b, _ = conv_module(
                branch_2, n_out_channel=256, filter_size=(3, 1), strides=(1, 1), padding='SAME', batch_norm_init=None,
                is_train=is_train, use_batchnorm=True, activation_fn='ReLU', name='Conv2d_0e_3x1'
            )

            branch_2 = tl.layers.ConcatLayer([branch_2a, branch_2b], concat_dim=3, name='concat_layer')

        with tf.variable_scope('Branch_3'):