import tensorflow as tf
from tensorflow.keras.preprocessing.image import ImageDataGenerator

# **Data Preparation (Emphasizing Quality)**

# Define paths to high-quality image datasets
train_dir = 'path/to/high_quality/train_images'
validation_dir = 'path/to/high_quality/validation_images'

# Data Augmentation to Increase Data Diversity and Robustness
train_datagen = ImageDataGenerator(
    rescale=1./255,
    rotation_range=20,
    width_shift_range=0.2,
    height_shift_range=0.2,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip=True,
    fill_mode='nearest'
)

validation_datagen = ImageDataGenerator(rescale=1./255)