Tusharsharma118

monitor = tf.keras.callbacks.EarlyStopping(monitor = 'val_loss',patience = 8,restore_best_weights = True, min_delta = 0.01)

history = model_regularized.fit_generator(train_generator, validation_data=validation_generator,steps_per_epoch=(len(train_split_images) / 32),epochs = 52,verbose=1,callbacks=[monitor])  # 32 = batch size

# Visualization code begins
training_loss = history.history['loss']
validation_loss = history.history['val_loss']

# Create count of the number of epochs

Tusharsharma118

monitor = tf.keras.callbacks.EarlyStopping(monitor = 'val_loss',patience = 8,restore_best_weights = True, min_delta = 0.01)

# Using the Callback

history = model_regularized.fit_generator(train_generator, validation_data=validation_generator,steps_per_epoch=(len(train_split_images) / 32),epochs = 52,verbose=1,callbacks=[monitor])

Tusharsharma118

training_datagen = tf.keras.preprocessing.image.ImageDataGenerator(zoom_range=0.3,width_shift_range=0.15,height_shift_range=0.15,shear_range=0.2,fill_mode='nearest')
validation_datagen = tf.keras.preprocessing.image.ImageDataGenerator()

# try and visualize your images
sign_generator = training_datagen.flow(train_images128[110:111], training_labels[110:111],batch_size=1)

sign = [next(sign_generator) for i in range(0,10)]
fig, ax = plt.subplots(1,10, figsize=(16, 6))
print('Labels:', [item[1][0] for item in sign])
l = [ax[i].imshow(sign[i][0][0]) for i in range(0,10)]

Tusharsharma118

def print_random_images(images, labels, number_of_images_to_display, color_map='hsv'):
   if number_of_images_to_display % 4 == 0:
       num_rows = number_of_images_to_display / 4
   else:
       num_rows = int(number_of_images_to_display / 4) + 1
   random_indices = [np.random.randint(0, images.shape[0]) for n in range(number_of_images_to_display)]
   # print(random_indices)
   for counter in range(number_of_images_to_display):
       index = random_indices[counter]
       plt.subplot(num_rows, 4, counter + 1)

Tusharsharma118

 {
    0: 'Warning for a bad road surface',
    1: 'Warning for a speed bump',
    2: 'Warning for a slippery road surface',
    3: 'Warning for a curve to the left',
    4: 'Warning for a curve to the right',
    5: 'Warning for a double curve, first left then right',  # Merge Classes 5 & 6 later
    6: 'Warning for a double curve, first left then right',
    7: 'Watch out for children ahead',
    8: 'Watch out for  cyclists',

Tusharsharma118

def load_images(data_directory):
  
  # lists to store Images and labels
  images = []
  labels = []
  log_index = 0
  # get list of all directories present in the data_directory path
  directories = [dir for dir in os.listdir(data_directory)
                 if os.path.isdir(os.path.join(data_directory,dir))] # to make sure that we include only directories and not any files present in the folder
  print(len(directories))

Tusharsharma118

print(os.getcwd())  # returns your current working directory
DATASET_PATH = """/My Drive/Ml Datasets/Belgian Img Traffic Signs Datasets"""  # Insert Data Path here relative to current working directory
os.chdir(DATASET_PATH)
print(os.getcwd()) # make sure you're in the directory containing dataset zipfiles 
# Extracting Training Data
with zipfile.ZipFile('BelgiumTSC_Training.zip', 'r') as zipObj:
  zipObj.extractall('TrainingData')
# Extracting Testing Data
with zipfile.ZipFile('BelgiumTSC_Testing.zip', 'r') as zipObj:
  zipObj.extractall('TestingData')