DV Dis

Member 2 – code - Generator Model
# Creating the Generator Model
KI = keras.initializers.RandomNormal(mean=0.0, stddev=0.02)
input_dim = 300
def Generator_Model():
Generator = Sequential()
Generator.add(Dense(8 * 8 * 512, input_dim=input_dim))
Generator.add(ReLU())
Generator.add(Reshape((8, 8, 512)))
Generator.add(Conv2DTranspose(256, (4, 4), strides=(2, 2), padding='same', kernel_initializer=KI,

activation='ReLU'))

activation='ReLU'))
activation='ReLU'))
Generator.add(Conv2D(3, (4, 4), padding='same', activation='sigmoid'))
return Generator
generator = Generator_Model()
generator.summary()
keras.utils.plot_model(generator, show_shapes=True)
…………………………………………………………………………………………………………………………………………………………
Member 3 code - Discriminator Model
# Creating the discriminator Model
def Discriminator_Model():
input_shape = (64, 64, 3)
discriminator = Sequential()
discriminator.add(Conv2D(64, kernel_size=(3, 3), activation='LeakyReLU',

input_shape=input_shape))
discriminator.add(MaxPooling2D(pool_size=(2, 2)))
discriminator.add(Conv2D(128, kernel_size=(3, 3), activation='LeakyReLU'))
discriminator.add(Conv2D(256, kernel_size=(3, 3), activation='LeakyReLU'))
discriminator.add(Flatten())
discriminator.add(Dense(256, activation='LeakyReLU'))
discriminator.add(Dense(1, activation='sigmoid'))
return discriminator
discriminator = Discriminator_Model()
discriminator.summary()
keras.utils.plot_model(discriminator, show_shapes=True)
……………………………………………………………………………………………………………………………………………………………
Member 4th - Training Loop and Model Evaluation
# DCGAN Model Training Step with Discriminator and Generator
class DCGAN(keras.Model):
# ...
# Definition of DCGAN class goes here
class DCGANMonitor(keras.callbacks.Callback):
# ...
# Definition of DCGANMonitor class goes here
# Training DCGAN on Image Dataset for 30 Epochs
epochs = 30
lr_g = 0.0003
lr_d = 0.0001
beta = 0.5
latent_dim = 300
dcgan = DCGAN(generator=generator, discriminator=discriminator, latent_dim=latent_dim)
dcgan.compile(g_optimizer=Adam(learning_rate=lr_g, beta_1=beta),
d_optimizer=Adam(learning_rate=lr_g, beta_1=beta), loss_fn=BinaryCrossentropy())
history = dcgan.fit(train_generator, epochs=epochs, callbacks=[DCGANMonitor()])
plt.figure(figsize=(40, 40))
# Generate and plot random images here
# Plot loss curves
def create_loss_figure(d_loss_values, g_loss_values):
# Function definition here
d_loss_values = history.history['d_loss']
g_loss_values = history.history['g_loss']
create_loss_figure(d_loss_values, g_loss_values)
…………………………………………………………………………………………………………………………………………………………
Member 1 – data handing and preprocessing
# Importing required libraries and tools
from tensorflow import keras
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
import seaborn as sns
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Dense, Flatten, Conv2DTranspose,

Reshape, BatchNormalization, Dropout, Input, ReLU, LeakyReLU
from keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.optimizers import Adam
from tensorflow.keras.losses import BinaryCrossentropy
from PIL import Image
import warnings
warnings.filterwarnings('ignore')
# Loading and Preparing Anime Face Images Dataset using Keras Image Data Generator
img_width, img_height = 256, 256
batchsize = 32
train = keras.utils.image_dataset_from_directory(
directory='/kaggle/input/anime-faces/data',
batch_size=batchsize,
image_size=(img_width, img_height))
# Visualizing a Batch of Anime Face Images

data_iterator = train.as_numpy_iterator()
batch = data_iterator.next()
fig, ax = plt.subplots(ncols=4, figsize=(10, 10))
for idx, img in enumerate(batch[0][:4]):
ax[idx].imshow(img.astype(int))
ax[idx].title.set_text(batch[1][idx])
# Generating Augmented Batches of Anime Face Images using ImageDataGenerator
DIR = '/kaggle/input/anime-faces/data' # path
train_datagen = ImageDataGenerator(rescale=1./255, horizontal_flip=True)
train_generator = train_datagen.flow_from_directory(
DIR,
target_size=(64, 64),
batch_size=batchsize,
class_mode=None)

DV Dis

Uploaded by

Copyright:

Available Formats

DV Dis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DV Dis

Uploaded by

Copyright:

Available Formats

Member 2 – code - Generator Model

# Creating the Generator Model

Generator.add(Dense(8 * 8 * 512, input_dim=input_dim))

Generator.add(Conv2DTranspose(256, (4, 4), strides=(2, 2), padding='same', kernel_initializer=KI,

Generator.add(Conv2DTranspose(128, (4, 4), strides=(2, 2), padding='same', kernel_initializer=KI,

Generator.add(Conv2D(3, (4, 4), padding='same', activation='sigmoid'))

Member 3 code - Discriminator Model

# Creating the discriminator Model

input_shape = (64, 64, 3)

discriminator.add(Conv2D(64, kernel_size=(3, 3), activation='LeakyReLU',

discriminator.add(Conv2D(128, kernel_size=(3, 3), activation='LeakyReLU'))

discriminator.add(Conv2D(256, kernel_size=(3, 3), activation='LeakyReLU'))

# DCGAN Model Training Step with Discriminator and Generator

# Definition of DCGAN class goes here

# Definition of DCGANMonitor class goes here

# Training DCGAN on Image Dataset for 30 Epochs

dcgan = DCGAN(generator=generator, discriminator=discriminator, latent_dim=latent_dim)

history = dcgan.fit(train_generator, epochs=epochs, callbacks=[DCGANMonitor()])

# Generate and plot random images here

# Plot loss curves

def create_loss_figure(d_loss_values, g_loss_values):

# Function definition here

Member 1 – data handing and preprocessing

# Importing required libraries and tools

from tensorflow import keras

import matplotlib.pyplot as plt

import seaborn as sns

from tensorflow.keras.models import Sequential

from tensorflow.keras.layers import Conv2D, MaxPooling2D, Dense, Flatten, Conv2DTranspose,

from keras.preprocessing.image import ImageDataGenerator

from tensorflow.keras.optimizers import Adam

from tensorflow.keras.losses import BinaryCrossentropy

from PIL import Image

img_width, img_height = 256, 256

# Visualizing a Batch of Anime Face Images

fig, ax = plt.subplots(ncols=4, figsize=(10, 10))

for idx, img in enumerate(batch[0][:4]):

# Generating Augmented Batches of Anime Face Images using ImageDataGenerator

DIR = '/kaggle/input/anime-faces/data' # path

train_datagen = ImageDataGenerator(rescale=1./255, horizontal_flip=True)

You might also like