Why the batch normalization layer is omitted in your implementation

[shanyucha]

In the original implementation of caffe-segnet, the bn layer is appended after conv layer. So why this implementation did not use batch normalization layer?

[Bonnie970]

SegNet-tensorflow/layers_object.py

Line 59 in 4f925b3

conv_out = tf.nn.relu(batch_norm(bias, is_training, scope))

There is BN wrapped in cnn layer.