Inteligencia Artificial - Como ampliar el conocimiento de mi red neuronal

import sys

import os

from PIL import Image

import matplotlib.pyplot as plt

import tensorflow as tf

from tensorflow.python.keras.preprocessing.image import ImageDataGenerator

from tensorflow.python.keras import optimizers, layers, models

from tensorflow.python.keras.models import Sequential

from tensorflow.python.keras.layers import Dropout, Flatten, Dense, Activation

from tensorflow.python.keras.layers import Convolution2D, MaxPooling2D

from tensorflow.python.keras import backend as K

K.clear_session()

data_train = './img/entrenamiento'

data_test = './img/test'

epochs = 20

steps_per_epoch = 1000

altura, longitud = 100, 100

batch_size = 32

steps_validation = 200

filtroConv1 = 32

filtroConv2 = 64

filtroConv1_size = (3,3)

filtroConv2_size = (2,2)

pool_size = (2,2)

tipos_imagen = 3

lr = 0.0005

train_labels = ['perro', 'gato', 'gorila']

test_labels = ['perro', 'gato', 'gorila']

entrenamiento_datagen = ImageDataGenerator(

    rescale = 1./255,

    shear_range = 0.2,

    zoom_range = 0.2,

    horizontal_flip = True

)

validacion_datagen = ImageDataGenerator(

    rescale = 1./255

)

train_images = entrenamiento_datagen.flow_from_directory(

    data_train,

    target_size = (altura, longitud),

    batch_size = batch_size,

    class_mode = 'categorical'

)

test_images = validacion_datagen.flow_from_directory(

    data_test,

    target_size = (altura, longitud),

    batch_size = batch_size,

    class_mode = 'categorical'

)

cnn = models.Sequential()

cnn.add(Convolution2D(filtroConv1, filtroConv1_size, padding = 'same', input_shape = (altura, longitud, 3), activation = 'relu'))

cnn.add(MaxPooling2D(pool_size = pool_size))

cnn.add(Convolution2D(filtroConv2, filtroConv2_size, padding = 'same'))

cnn.add(MaxPooling2D(pool_size = pool_size))

cnn.add(Flatten())

cnn.add(Dense(256, activation = 'relu'))

cnn.add(Dropout(0.5))

cnn.add(Dense(tipos_imagen, activation = 'softmax'))

cnn.compile(loss = 'categorical_crossentropy', optimizer = tf.keras.optimizers.Adam(learning_rate = lr), metrics = ['accuracy'])

history = cnn.fit(train_images, epochs = epochs, validation_data = test_images)

print(train_images.class_indices)

dir = './model/'

if not os.path.exists(dir):

    os.mkdir(dir)

cnn.save('./model/model.h5')

cnn.save_weights('./model/pesos.h5')