Upgrade Cyclegan to Keras 3 (with Tensorflow backend) (#1944)

* upgrade cyclegan.py to Keras 3 with tensorflow backend

* enabling keras 3

* Increase training epochs

* add ipynb, md files and images

* remove training logs

Author: pksX01

examples/generative/cyclegan.py

@@ -2,7 +2,7 @@
 Title: CycleGAN
 Author: [A_K_Nain](https://twitter.com/A_K_Nain)
 Date created: 2020/08/12
-Last modified: 2020/08/12
+Last modified: 2024/09/30
 Description: Implementation of CycleGAN.
 Accelerator: GPU
 """
@@ -17,26 +17,26 @@
 CycleGAN tries to learn this mapping without requiring paired input-output images,
 using cycle-consistent adversarial networks.
 
-- [Paper](https://arxiv.org/pdf/1703.10593.pdf)
+- [Paper](https://arxiv.org/abs/1703.10593)
 - [Original implementation](https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix)
 """
 
 """
 ## Setup
 """
 
-
+import os
 import numpy as np
 import matplotlib.pyplot as plt
 import tensorflow as tf
-from tensorflow import keras
-from tensorflow.keras import layers
-import tensorflow_addons as tfa
+import keras
+from keras import layers, ops
 import tensorflow_datasets as tfds
 
 tfds.disable_progress_bar()
 autotune = tf.data.AUTOTUNE
 
+os.environ["KERAS_BACKEND"] = "tensorflow"
 
 """
 ## Prepare the dataset
@@ -47,7 +47,7 @@
 """
 
 # Load the horse-zebra dataset using tensorflow-datasets.
-dataset, _ = tfds.load("cycle_gan/horse2zebra", with_info=True, as_supervised=True)
+dataset, _ = tfds.load(name="cycle_gan/horse2zebra", with_info=True, as_supervised=True)
 train_horses, train_zebras = dataset["trainA"], dataset["trainB"]
 test_horses, test_zebras = dataset["testA"], dataset["testB"]
 
@@ -65,7 +65,7 @@
 
 
 def normalize_img(img):
-    img = tf.cast(img, dtype=tf.float32)
+    img = ops.cast(img, dtype=tf.float32)
     # Map values in the range [-1, 1]
     return (img / 127.5) - 1.0
 
@@ -74,7 +74,7 @@ def preprocess_train_image(img, label):
     # Random flip
     img = tf.image.random_flip_left_right(img)
     # Resize to the original size first
-    img = tf.image.resize(img, [*orig_img_size])
+    img = ops.image.resize(img, [*orig_img_size])
     # Random crop to 256X256
     img = tf.image.random_crop(img, size=[*input_img_size])
     # Normalize the pixel values in the range [-1, 1]
@@ -84,7 +84,7 @@ def preprocess_train_image(img, label):
 
 def preprocess_test_image(img, label):
     # Only resizing and normalization for the test images.
-    img = tf.image.resize(img, [input_img_size[0], input_img_size[1]])
+    img = ops.image.resize(img, [input_img_size[0], input_img_size[1]])
     img = normalize_img(img)
     return img
 
@@ -165,7 +165,7 @@ def call(self, input_tensor, mask=None):
             [padding_width, padding_width],
             [0, 0],
         ]
-        return tf.pad(input_tensor, padding_tensor, mode="REFLECT")
+        return ops.pad(input_tensor, padding_tensor, mode="REFLECT")
 
 
 def residual_block(
@@ -190,7 +190,9 @@ def residual_block(
         padding=padding,
         use_bias=use_bias,
     )(x)
-    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
+    x = keras.layers.GroupNormalization(groups=1, gamma_initializer=gamma_initializer)(
+        x
+    )
     x = activation(x)
 
     x = ReflectionPadding2D()(x)
@@ -202,7 +204,9 @@ def residual_block(
         padding=padding,
         use_bias=use_bias,
     )(x)
-    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
+    x = keras.layers.GroupNormalization(groups=1, gamma_initializer=gamma_initializer)(
+        x
+    )
     x = layers.add([input_tensor, x])
     return x
 
@@ -226,7 +230,9 @@ def downsample(
         padding=padding,
         use_bias=use_bias,
     )(x)
-    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
+    x = keras.layers.GroupNormalization(groups=1, gamma_initializer=gamma_initializer)(
+        x
+    )
     if activation:
         x = activation(x)
     return x
@@ -251,7 +257,9 @@ def upsample(
         kernel_initializer=kernel_initializer,
         use_bias=use_bias,
     )(x)
-    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
+    x = keras.layers.GroupNormalization(groups=1, gamma_initializer=gamma_initializer)(
+        x
+    )
     if activation:
         x = activation(x)
     return x
@@ -298,7 +306,9 @@ def get_resnet_generator(
     x = layers.Conv2D(filters, (7, 7), kernel_initializer=kernel_init, use_bias=False)(
         x
     )
-    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
+    x = keras.layers.GroupNormalization(groups=1, gamma_initializer=gamma_initializer)(
+        x
+    )
     x = layers.Activation("relu")(x)
 
     # Downsampling
@@ -581,14 +591,14 @@ def on_epoch_end(self, epoch, logs=None):
 
 
 def generator_loss_fn(fake):
-    fake_loss = adv_loss_fn(tf.ones_like(fake), fake)
+    fake_loss = adv_loss_fn(ops.ones_like(fake), fake)
     return fake_loss
 
 
 # Define the loss function for the discriminators
 def discriminator_loss_fn(real, fake):
-    real_loss = adv_loss_fn(tf.ones_like(real), real)
-    fake_loss = adv_loss_fn(tf.zeros_like(fake), fake)
+    real_loss = adv_loss_fn(ops.ones_like(real), real)
+    fake_loss = adv_loss_fn(ops.zeros_like(fake), fake)
     return (real_loss + fake_loss) * 0.5
 
 
@@ -599,16 +609,16 @@ def discriminator_loss_fn(real, fake):
 
 # Compile the model
 cycle_gan_model.compile(
-    gen_G_optimizer=keras.optimizers.legacy.Adam(learning_rate=2e-4, beta_1=0.5),
-    gen_F_optimizer=keras.optimizers.legacy.Adam(learning_rate=2e-4, beta_1=0.5),
-    disc_X_optimizer=keras.optimizers.legacy.Adam(learning_rate=2e-4, beta_1=0.5),
-    disc_Y_optimizer=keras.optimizers.legacy.Adam(learning_rate=2e-4, beta_1=0.5),
+    gen_G_optimizer=keras.optimizers.Adam(learning_rate=2e-4, beta_1=0.5),
+    gen_F_optimizer=keras.optimizers.Adam(learning_rate=2e-4, beta_1=0.5),
+    disc_X_optimizer=keras.optimizers.Adam(learning_rate=2e-4, beta_1=0.5),
+    disc_Y_optimizer=keras.optimizers.Adam(learning_rate=2e-4, beta_1=0.5),
     gen_loss_fn=generator_loss_fn,
     disc_loss_fn=discriminator_loss_fn,
 )
 # Callbacks
 plotter = GANMonitor()
-checkpoint_filepath = "./model_checkpoints/cyclegan_checkpoints.{epoch:03d}"
+checkpoint_filepath = "./model_checkpoints/cyclegan_checkpoints.weights.h5"
 model_checkpoint_callback = keras.callbacks.ModelCheckpoint(
     filepath=checkpoint_filepath, save_weights_only=True
 )
@@ -617,31 +627,20 @@ def discriminator_loss_fn(real, fake):
 # 7 minutes on a single P100 backed machine.
 cycle_gan_model.fit(
     tf.data.Dataset.zip((train_horses, train_zebras)),
-    epochs=1,
+    epochs=90,
     callbacks=[plotter, model_checkpoint_callback],
 )
 
 """
 Test the performance of the model.
-
-You can use the trained model hosted on [Hugging Face Hub](https://huggingface.co/keras-io/CycleGAN)
-and try the demo on [Hugging Face Spaces](https://huggingface.co/spaces/keras-io/CycleGAN).
 """
 
 
-# This model was trained for 90 epochs. We will be loading those weights
-# here. Once the weights are loaded, we will take a few samples from the test
-# data and check the model's performance.
-
-"""shell
-curl -LO https://github.com/AakashKumarNain/CycleGAN_TF2/releases/download/v1.0/saved_checkpoints.zip
-unzip -qq saved_checkpoints.zip
-"""
+# Once the weights are loaded, we will take a few samples from the test data and check the model's performance.
 
 
 # Load the checkpoints
-weight_file = "./saved_checkpoints/cyclegan_checkpoints.090"
-cycle_gan_model.load_weights(weight_file).expect_partial()
+cycle_gan_model.load_weights(checkpoint_filepath)
 print("Weights loaded successfully")
 
 _, ax = plt.subplots(4, 2, figsize=(10, 15))