extra_training.py

import os
import pickle
import numpy as np 

os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2"

import tensorflow as tf
from tensorflow import keras
from keras import layers

from time import perf_counter
from sklearn.model_selection import train_test_split

list_of_images = 'list_of_images.pkl'
list_of_answers = 'list_of_answers.pkl'

open_images_file = open(list_of_images, "rb")
images_list = pickle.load(open_images_file)
open_images_file.close()

open_answers_file = open(list_of_answers, "rb")
answers_list = pickle.load(open_answers_file)
open_answers_file.close()

images_list = list(images_list)
answers_list = list(answers_list)

images_list = np.array(images_list)
answers_list = np.array(answers_list)

train_images, test_images, train_labels, test_labels = train_test_split(
    images_list, answers_list, test_size=0.25
)

train_images = train_images  / 255.0
test_images = test_images / 255.0

print(train_images.shape)
print(train_labels.shape)

model = tf.keras.models.load_model("BCM_LiteTurboPrime4.h5")

print(model.summary())

loss = keras.losses.BinaryCrossentropy(from_logits=True)
optim = keras.optimizers.Adam(learning_rate=0.01)
metrics = ["accuracy"]

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

batch_size = 64
epochs = 3

start_time = perf_counter()

model.fit(train_images, train_labels, epochs=epochs, batch_size=batch_size, verbose=2)
model.evaluate(test_images, test_labels, batch_size=batch_size, verbose=2)

model.save("BCM_LiteTurboPrime5.h5")

end_time = perf_counter()

print("\n-----------------------------\n")

print(f"Operation took {round(end_time - start_time)} second(s) to complete!")