Merge branch 'master' of skozl.com:e4-gan

3 files changed, 280 insertions, 17 deletions

diff --git a/cdcgan.py b/cdcgan.py
index 8d59a03..effc89b 100755
--- a/cdcgan.py
+++ b/cdcgan.py
@@ -7,7 +7,7 @@ from keras.datasets import mnist
 from keras.layers import Input, Dense, Reshape, Flatten, Dropout, multiply
 from keras.layers import BatchNormalization, Embedding, Activation, ZeroPadding2D
 from keras.layers import LeakyReLU
-from keras.layers import UpSampling2D, Conv2D
+from keras.layers import UpSampling2D, Conv2D, Conv2DTranspose
 from keras.models import Sequential, Model
 from keras.optimizers import Adam
 
@@ -65,25 +65,22 @@ class CDCGAN():
 
         model.add(Dense(128 * 7 * 7, activation="relu", input_dim=self.latent_dim))
         model.add(Reshape((7, 7, 128)))
-        model.add(UpSampling2D())
 
-        for i in range(self.conv_layers):
-            model.add(Conv2D(128, kernel_size=3, padding="same"))
-            model.add(BatchNormalization())
-            model.add(Activation("relu"))
-
-        model.add(UpSampling2D())
+        model.add(Conv2DTranspose(256, kernel_size=3, padding="same", strides=(2,2)))
+        model.add(BatchNormalization())
+        model.add(Activation("relu"))
 
-        for i in range(self.conv_layers):
-            model.add(Conv2D(64, kernel_size=3, padding="same"))
-            model.add(BatchNormalization())
+        model.add(Conv2DTranspose(128, kernel_size=3, padding="same", strides=(2,2)))
+        model.add(BatchNormalization())
+        model.add(Activation("relu"))
 
-            model.add(Activation("relu"))
+        model.add(Conv2DTranspose(64, kernel_size=3, padding="same"))
+        model.add(BatchNormalization())
+        model.add(Activation("relu"))
 
-        model.add(Conv2D(self.channels, kernel_size=3, padding="same"))
+        model.add(Conv2DTranspose(1, kernel_size=3, padding="same"))
         model.add(Activation("tanh"))
 
-        #model.summary()
 
         noise = Input(shape=(self.latent_dim,))
         label = Input(shape=(1,), dtype='int32')
@@ -91,6 +88,8 @@ class CDCGAN():
         model_input = multiply([noise, label_embedding])
         img = model(model_input)
 
+        #model.summary()
+
         return Model([noise, label], img)
 
     def build_discriminator(self):
diff --git a/dcgan.py b/dcgan.py
index 4317994..a362f69 100644
--- a/dcgan.py
+++ b/dcgan.py
@@ -1,4 +1,5 @@
 from __future__ import print_function, division
+import tensorflow.keras as keras
 from keras.datasets import mnist
 from keras.layers import Input, Dense, Reshape, Flatten, Dropout
 from keras.layers import BatchNormalization, Activation, ZeroPadding2D
@@ -76,7 +77,7 @@ class DCGAN():
         model.add(Conv2D(self.channels, kernel_size=3, padding="same"))
         model.add(Activation("tanh"))
 
-        #model.summary()
+        model.summary()
 
         noise = Input(shape=(self.latent_dim,))
         img = model(noise)
@@ -191,8 +192,6 @@ class DCGAN():
         fig.savefig("images/mnist_%d.png" % epoch)
         plt.close()
 
-'''
 if __name__ == '__main__':
     dcgan = DCGAN()
     dcgan.train(epochs=4000, batch_size=32, save_interval=50)
-'''
diff --git a/ncdcgan.py b/ncdcgan.py
new file mode 100755
index 0000000..ccb99d3
--- /dev/null
+++ b/ncdcgan.py
@@ -0,0 +1,265 @@
+from __future__ import print_function, division
+import tensorflow as keras
+
+import tensorflow as tf
+import tensorflow.keras as keras
+from keras.datasets import mnist
+from keras.layers import Input, Dense, Reshape, Flatten, Dropout, Multiply
+from keras.layers import BatchNormalization, Embedding, Activation, ZeroPadding2D
+from keras.layers import LeakyReLU
+from keras.layers import UpSampling2D, Conv2D, Conv2DTranspose
+from keras.models import Sequential, Model
+from keras.optimizers import Adam
+
+import matplotlib.pyplot as plt
+import matplotlib.gridspec as gridspec
+
+from tqdm import tqdm
+
+import sys
+
+import numpy as np
+
+class nCDCGAN():
+    def __init__(self, conv_layers = 1, num_classes = 10):
+        # Input shape
+        self.num_classes = num_classes
+        self.img_rows = 28
+        self.img_cols = 28
+        self.channels = 1
+        self.img_shape = (self.img_rows, self.img_cols, self.channels)
+        self.latent_dim = 100
+        self.conv_layers = conv_layers
+
+        optimizer = Adam(0.002, 0.5)
+
+        noise = Input(shape=(self.latent_dim,))
+        label = Input(shape=(1,))
+
+        # Build the generator
+        self.generator = self.build_generator(noise, label)
+
+        ph_img = Input(shape=self.img_shape)
+
+        # Build and compile the discriminator
+        self.discriminator = self.build_discriminator(ph_img, label)
+        self.discriminator.compile(loss='binary_crossentropy',
+            optimizer=optimizer,
+            metrics=['accuracy'])
+
+        img = self.generator([noise, label])
+
+        
+
+        # For the combined model we will only train the generator
+        self.discriminator.trainable = False
+
+        # The discriminator takes generated images as input and determines validity
+        valid = self.discriminator([img, label])
+
+
+        # The combined model  (stacked generator and discriminator)
+        # Trains generator to fool discriminator
+        self.combined = Model([noise, label], valid)
+        self.combined.compile(loss=['binary_crossentropy'],
+            optimizer=optimizer)
+
+
+    def build_generator(self, noise, con):
+
+        n_channel = 64
+        kernel_size = 3
+
+        con1 = Dense(n_channel, activation='tanh')(con) #model settings
+        con1 = Reshape((1,1,n_channel))(con1)
+        con1 = UpSampling2D((28,28))(con1)
+
+        hid = Dense(n_channel*7*7, activation='relu')(noise)
+        hid = Reshape((7,7,n_channel))(hid)
+
+        hid = Conv2DTranspose(n_channel, kernel_size=kernel_size, strides=2, padding="same")(hid)
+        hid = BatchNormalization(momentum=0.8)(hid)
+        hid = Activation("relu")(hid)
+
+        hid = Conv2DTranspose(n_channel, kernel_size=kernel_size, strides=2, padding="same")(hid)
+        hid = BatchNormalization(momentum=0.8)(hid)
+        hid = Activation("relu")(hid) # -> 128x144x144
+        hid = Multiply()([hid, con1])
+
+        hid = Conv2D(n_channel, kernel_size=kernel_size, strides=1, padding="same")(hid)
+        hid = BatchNormalization(momentum=0.8)(hid)
+        hid = Activation("relu")(hid) # -> 128x144x144
+        hid = Multiply()([hid, con1])
+
+        hid = Conv2D(n_channel, kernel_size=kernel_size, strides=1, padding="same")(hid)
+        hid = BatchNormalization(momentum=0.8)(hid)
+        hid = Activation("relu")(hid) # -> 128x144x144
+        hid = Multiply()([hid, con1])
+
+        hid = Conv2D(1, kernel_size=kernel_size, strides=1, padding="same")(hid)
+        out = Activation("tanh")(hid)
+
+        model =  Model([noise, con], out)
+        model.summary()
+        return model
+
+
+    def build_discriminator(self, img, con):
+
+        n_channel = 64
+        kernel_size = 3
+
+        con1 = Dense(n_channel, activation='tanh')(con) #model settings
+        con1 = Reshape((1,1,n_channel))(con1)
+        con1 = UpSampling2D((28,28))(con1)
+
+
+        hid = Conv2D(n_channel, kernel_size=kernel_size, strides=1, padding="same")(img)
+        hid = BatchNormalization(momentum=0.8)(hid)
+        hid = LeakyReLU(alpha=0.2)(hid) # -> 32
+        hid = Multiply()([hid, con1]) # -> 128x128xn_channel
+
+        hid = Conv2D(n_channel, kernel_size=kernel_size, strides=1, padding="same")(hid)
+        hid = BatchNormalization(momentum=0.8)(hid)
+        hid = LeakyReLU(alpha=0.2)(hid) # -> 32
+        hid = Multiply()([hid, con1])
+
+        hid = Conv2D(n_channel, kernel_size=kernel_size, strides=1, padding="same")(hid)
+        hid = BatchNormalization(momentum=0.8)(hid)
+        hid = LeakyReLU(alpha=0.2)(hid) # -> 32
+        hid = Multiply()([hid, con1])
+
+
+        hid = Conv2D(n_channel, kernel_size=kernel_size, strides=2, padding="same")(hid)
+        hid = BatchNormalization(momentum=0.8)(hid)
+        hid = LeakyReLU(alpha=0.2)(hid) # -> 64
+
+        hid = Conv2D(n_channel, kernel_size=kernel_size, strides=2, padding="same")(hid)
+        hid = BatchNormalization(momentum=0.8)(hid)
+        hid = LeakyReLU(alpha=0.2)(hid) # -> 32
+
+        hid = Flatten()(hid)
+
+        hid = Dropout(0.1)(hid)
+
+        out = Dense(1, activation='sigmoid')(hid)
+
+        model = Model(inputs=[img, con], outputs=out)
+        model.summary()
+        return model
+
+    def train(self, epochs, batch_size=128, sample_interval=50, graph=False, smooth_real=1, smooth_fake=0):
+
+        # Load the dataset
+        (X_train, y_train), (_, _) = mnist.load_data()
+
+        # Configure input
+        X_train = (X_train.astype(np.float32) - 127.5) / 127.5
+        X_train = np.expand_dims(X_train, axis=3)
+        y_train = y_train.reshape(-1, 1)
+
+        # Adversarial ground truths
+        valid = np.ones((batch_size, 1))
+        fake = np.zeros((batch_size, 1))
+        
+        xaxis = np.arange(epochs)
+        loss = np.zeros((2,epochs))
+        for epoch in tqdm(range(epochs)):
+
+            # ---------------------
+            #  Train Discriminator
+            # ---------------------
+
+            # Select a random half batch of images
+            idx = np.random.randint(0, X_train.shape[0], batch_size)
+            imgs, labels = X_train[idx], y_train[idx]
+
+            # Sample noise as generator input
+            noise = np.random.normal(0, 1, (batch_size, 100))
+
+            # Generate a half batch of new images
+            gen_imgs = self.generator.predict([noise, labels])
+
+            # Train the discriminator
+            d_loss_real = self.discriminator.train_on_batch([imgs, labels], valid*smooth_real)
+            d_loss_fake = self.discriminator.train_on_batch([gen_imgs, labels], valid*smooth_fake)
+            d_loss = 0.5 * np.add(d_loss_real, d_loss_fake)
+
+            # ---------------------
+            #  Train Generator
+            # ---------------------
+
+            # Condition on labels
+            sampled_labels = np.random.randint(0, 10, batch_size).reshape(-1, 1)
+            # Train the generator
+            g_loss = self.combined.train_on_batch([noise, sampled_labels], valid)
+
+            # Plot the progress
+            #print ("%d [D loss: %f, acc.: %.2f%%] [G loss: %f]" % (epoch, d_loss[0], 100*d_loss[1], g_loss))
+            loss[0][epoch] = d_loss[0]
+            loss[1][epoch] = g_loss
+
+            # If at save interval => save generated image samples
+            if epoch % sample_interval == 0:
+                self.sample_images(epoch)
+            
+        if graph:
+          plt.plot(xaxis,loss[0])
+          plt.plot(xaxis,loss[1])
+          plt.legend(('Discriminator', 'Generator'), loc='best')
+          plt.xlabel('Epoch')
+          plt.ylabel('Binary Crossentropy Loss')
+
+    def sample_images(self, epoch):
+        r, c = 2, 5
+        noise = np.random.normal(0, 1, (r * c, 100))
+        sampled_labels = np.arange(0, 10).reshape(-1, 1)
+
+        #using dummy_labels would just print zeros to help identify image quality
+        #dummy_labels = np.zeros(32).reshape(-1, 1)
+        
+        gen_imgs = self.generator.predict([noise, sampled_labels])
+
+        # Rescale images 0 - 1
+        gen_imgs = 0.5 * gen_imgs + 0.5
+
+        fig, axs = plt.subplots(r, c)
+        cnt = 0
+        for i in range(r):
+            for j in range(c):
+                axs[i,j].imshow(gen_imgs[cnt,:,:,0], cmap='gray')
+                axs[i,j].set_title("Digit: %d" % sampled_labels[cnt])
+                axs[i,j].axis('off')
+                cnt += 1
+        fig.savefig("images/%d.png" % epoch)
+        plt.close()
+        
+    def generate_data(self):
+      noise_train = np.random.normal(0, 1, (55000, 100))
+      noise_test = np.random.normal(0, 1, (10000, 100))
+      noise_val = np.random.normal(0, 1, (5000, 100))
+
+      labels_train = np.zeros(55000).reshape(-1, 1)
+      labels_test = np.zeros(10000).reshape(-1, 1)
+      labels_val = np.zeros(5000).reshape(-1, 1)
+
+      for i in range(10):
+        labels_train[i*5500:-1] = i
+        labels_test[i*1000:-1] = i
+        labels_val[i*500:-1] = i
+
+      train_data = self.generator.predict([noise_train, labels_train])
+      test_data = self.generator.predict([noise_test, labels_test])
+      val_data = self.generator.predict([noise_val, labels_val])
+
+      labels_train = keras.utils.to_categorical(labels_train, 10)
+      labels_test = keras.utils.to_categorical(labels_test, 10)
+      labels_val = keras.utils.to_categorical(labels_val, 10)
+
+      return train_data, test_data, val_data, labels_train, labels_test, labels_val
+
+'''
+if __name__ == '__main__':
+    cdcgan = nCDCGAN()
+    cdcgan.train(epochs=4000, batch_size=32)
+'''