Simple Tensorflow 2.0 implementation of Batch-Instance Normalization (NIPS 2018).
Originally implemented in Tensorflow version 1 by taki0112.
Edited by Seunggeon Lim in reference to InstanceNormalization layer from Tensorflow examples pix2pix.py

Sequential API

model = tf.keras.Sequential()
model.add(BatchInstanceNormalization())

Functional API

conv = tf.keras.layers.Conv2D()(input)
norm = BatchInstanceNormalization()(conv)