def build_discriminator(self, k=4):
inputs = Input(shape=self.input_shape)
x = inputs
x = conv(f=128, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = conv(f=256, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = conv(f=256, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = conv(f=512, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
f = Flatten()(x)
x = Dense(1024, kernel_regularizer=l2(0.001))(f)
x = LeakyReLU(0.2)(x)
x = Dense(1, activation='sigmoid', kernel_regularizer=l2(0.001))(x)
return Model(inputs, [x, f])
def build_classifier(self, k=4):
inputs = Input(shape=self.input_shape)
x = inputs
x = conv(f=128, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = conv(f=256, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = conv(f=256, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = conv(f=512, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
f = Flatten()(x)
x = Dense(1024, kernel_regularizer=l2(0.001))(f)
x = LeakyReLU(0.2)(x)
x = Dense(self.num_attrs,
activation='softmax',
kernel_regularizer=l2(0.001))(x)
return Model(inputs, [x, f])
def build_encoder(self, output_dims, k=4):
x_inputs = Input(shape=self.input_shape)
c_inputs = Input(shape=(self.num_attrs, ))
c = Reshape((1, 1, self.num_attrs))(c_inputs)
c = UpSampling2D(size=self.input_shape[:2])(c)
x = Concatenate(axis=-1)([x_inputs, c])
x = conv(f=128, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = conv(f=256, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = conv(f=256, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = conv(f=512, k=k, stride=2)(x)
x = LeakyReLU(0.2)(x)
x = Flatten()(x)
x = Dense(1024, kernel_regularizer=l2(0.001))(x)
x = LeakyReLU(0.2)(x)
x = Dense(output_dims, kernel_regularizer=l2(0.001))(x)
return Model([x_inputs, c_inputs], x)
def __init__(self, conv_dim=64, noise_dim=32, init_zero_weights=False):
super(CycleGenerator, self).__init__()
self.conv_dim = conv_dim
self.noise_dim = noise_dim
# 1. Define the encoder part of the generator (that extracts features from the input image)
self.conv1 = conv(3, conv_dim, 4, padding=1, stride=2)
self.conv2 = conv(conv_dim, conv_dim * 2, 4, padding=1, stride=2)
# 2. Define the transformation part of the generator
self.resnet_block = ResnetBlock(conv_dim * 2)
# 3. Define the decoder part of the generator (that builds up the output image from features)
self.deconv1 = deconv(conv_dim * 2 + noise_dim,
conv_dim,
4,
padding=1,
stride=2)
self.deconv2 = deconv(conv_dim,
3,
4,
padding=1,
stride=2,
batch_norm=False)
from models import conv, xception
import numpy as np
import time
import tensorflow as tf
from tensorflow.keras.callbacks import TensorBoard
tensorboard_cb = TensorBoard(log_dir="logs/", profile_batch=0)
WIDTH = 80
HEIGHT = 60
LEARNING_RATE = 0.001
EPOCHS = 100
BATCH_SIZE = 75
MODEL_NAME = f"SelfDrivingCar-{time.time()}.h5"
model = conv(WIDTH, HEIGHT, LEARNING_RATE)
train_data = np.load("balanced_data.npy", allow_pickle=True)
train = train_data[:-100]
test = train_data[-100:]
x_train = np.array([sample[0] for sample in train],
dtype="int32").reshape(-1, WIDTH, HEIGHT, 1)
y_train = np.array([sample[1] for sample in train], dtype="int32")
x_test = np.array([sample[0] for sample in test],
dtype="int32").reshape(-1, WIDTH, HEIGHT, 1)
y_test = np.array([sample[1] for sample in test], dtype="int32")
print(x_train)