train.py

import cv2
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Dense, Dropout, Flatten
from keras.optimizers import Adam
from keras.preprocessing.image import ImageDataGenerator
from keras.callbacks import LearningRateScheduler, EarlyStopping

# Initialize image data generator with rescaling and data augmentation
train_data_gen = ImageDataGenerator(rescale=1.0/255)
validation_data_gen = ImageDataGenerator(rescale=1.0/255)

# Preprocess all train images
train_generator = train_data_gen.flow_from_directory(
    'data/train',
    target_size=(48, 48),
    batch_size=64,
    color_mode="grayscale",
    class_mode='categorical'
)

# Preprocess all validation images
validation_generator = validation_data_gen.flow_from_directory(
    'data/test',
    target_size=(48, 48),
    batch_size=64,
    color_mode="grayscale",
    class_mode='categorical'
)

# create model structure
emotion_model = Sequential()

emotion_model.add(Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=(48, 48, 1)))
emotion_model.add(Conv2D(64, kernel_size=(3, 3), activation='relu'))
emotion_model.add(MaxPooling2D(pool_size=(2, 2)))
emotion_model.add(Dropout(0.25))

emotion_model.add(Conv2D(128, kernel_size=(3, 3), activation='relu'))
emotion_model.add(MaxPooling2D(pool_size=(2, 2)))
emotion_model.add(Conv2D(128, kernel_size=(3, 3), activation='relu'))
emotion_model.add(MaxPooling2D(pool_size=(2, 2)))
emotion_model.add(Dropout(0.25))

emotion_model.add(Flatten())
emotion_model.add(Dense(1024, activation='relu'))
emotion_model.add(Dropout(0.5))
emotion_model.add(Dense(7, activation='softmax'))

cv2.ocl.setUseOpenCL(False)

emotion_model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

# Learning Rate Schedule
def lr_schedule(epoch):
    lr = 0.0001
    if epoch > 30:
        lr *= 0.1
    elif epoch > 20:
        lr *= 0.5
    return lr

lr_scheduler = LearningRateScheduler(lr_schedule)

# Early Stopping
early_stopping = EarlyStopping(monitor='val_loss', patience=10, restore_best_weights=True)

# Train the neural network/model
emotion_model_info = emotion_model.fit(
    train_generator,
    steps_per_epoch=28709 // 64,
    epochs=50,
    validation_data=validation_generator,
    validation_steps=7178 // 64,
    callbacks=[lr_scheduler, early_stopping]
)

# Save model structure in JSON file
model_json = emotion_model.to_json()
with open("emotion_model.json", "w") as json_file:
    json_file.write(model_json)

# Save trained model weights in .h5 file
emotion_model.save_weights('emotion_model.h5')