dermy-models/compile.py

import tensorflow as tf
import tensorflow.keras as tfk
import os


# Parameters
DATA_DIR = os.path.join(os.getcwd(), "data")
IMG_SIZE = (224, 224)
IMG_SHAPE = IMG_SIZE + (3,)
BATCH_SIZE = 64
AUTOTUNE = tf.data.AUTOTUNE
BASE_LEARNING_RATE = 0.0001
CLASSES = 2


# Import Data
training_path = os.path.join(DATA_DIR, "train")
test_path = os.path.join(DATA_DIR, "test")


training_data = tfk.utils.image_dataset_from_directory(training_path,
                                                       shuffle=True,
                                                       batch_size=BATCH_SIZE,
                                                       image_size=IMG_SIZE,
                                                       validation_split=0.2,
                                                       subset="training",
                                                       seed=1234)
validation_data = tfk.utils.image_dataset_from_directory(training_path,
                                                         shuffle=True,
                                                         batch_size=BATCH_SIZE,
                                                         image_size=IMG_SIZE,
                                                         validation_split=0.2,
                                                         subset="validation",
                                                         seed=1234)
test_data = tfk.utils.image_dataset_from_directory(test_path,
                                                   shuffle=True,
                                                   batch_size=BATCH_SIZE,
                                                   image_size=IMG_SIZE)


# Init Prefetching
training_data = training_data.prefetch(buffer_size=AUTOTUNE)
validation_data = validation_data.prefetch(buffer_size=AUTOTUNE)
test_data = test_data.prefetch(buffer_size=AUTOTUNE)


# Data Augmentation Layer
data_augmentation = tf.keras.Sequential([
    tf.keras.layers.RandomFlip('horizontal'),
    tf.keras.layers.RandomRotation(0.2)
    ])


# Create Base Model From MobileNetV3Large
base_model = tf.keras.applications.MobileNetV3Large(
        input_shape=IMG_SHAPE,
        include_top=False,
        weights="imagenet"
        )

image_batch, label_batch = next(iter(training_data))
feature_batch = base_model(image_batch)

base_model.trainable = False


# Add Classification Header
global_avg_layer = tf.keras.layers.GlobalAveragePooling2D()
feature_batch_avg = global_avg_layer(feature_batch)

prediction_layer = tf.keras.layers.Dense(CLASSES, activation="softmax")
predication_batch = prediction_layer(feature_batch_avg)

inputs = tf.keras.Input(shape=IMG_SHAPE)
x = data_augmentation(inputs)
x = base_model(x, training=False)
x = global_avg_layer(x)
x = tf.keras.layers.Dropout(0.2)(x)
outputs = prediction_layer(x)

model = tf.keras.Model(inputs, outputs)


# Hotencode Data
training_data = training_data.map(lambda x, y: (x, tf.one_hot(y, CLASSES)))
validation_data = validation_data.map(lambda x, y: (x, tf.one_hot(y, CLASSES)))
test_data = test_data.map(lambda x, y: (x, tf.one_hot(y, CLASSES)))


# Compile Model
optimizer = tf.keras.optimizers.Adam(learning_rate=BASE_LEARNING_RATE)
loss = tf.keras.losses.CategoricalCrossentropy()
metrics = [tf.keras.metrics.CategoricalAccuracy()]

model.compile(optimizer=optimizer, loss=loss, metrics=metrics)


# Train the Model
initial_epochs = 50

loss0, accuracy0 = model.evaluate(validation_data)

print(f"initial loss: {loss0}")
print(f"initial accuracy: {accuracy0}")

lr_schedule = tf.keras.callbacks.ReduceLROnPlateau(
        monitor="val_loss",
        factor=0.1,
        patience=5,
        min_lr=1e-6
        )

early_stopping = tf.keras.callbacks.EarlyStopping(
        monitor="val_loss",
        patience=10,
        restore_best_weights=True
        )

history = model.fit(training_data,
                    epochs=initial_epochs,
                    validation_data=validation_data,
                    callbacks=[lr_schedule, early_stopping])


# Evaluate Model
results = model.evaluate(validation_data)
print(f"Validation Loss: {results[0]}")
print(f"Validation Accuracy: {results[1]}")

results = model.evaluate(test_data)
print(f"Test Loss: {results[0]}")
print(f"Test Accuracy: {results[1]}")


# Save Model
model.save("models/mobilenet_v3.keras")