Curso de Modelos Predictivos con Python

Optimizadores

Optimizador Momentum
Algoritmo de gradiente adaptativo
Adadelta

Adadelta

Elimina la nececidad de seleccionar una tasa manual

Arquitecturas CNN

Combinan capas

Input > Convolution > Pooling > Convolution > Pooling > Fully connected

Tema 1: Clasificador de imagenes rurales vs ciudad

Entrenamiento

Para este curso estaremos utilizando las siguientes librerías:

from keras.models import Sequential, Model
from keras.layers import Conv2D, MaxPool2D, Dense, Flatten, Dropout, BatchNormalization, Input
from keras.callbacks import TensorBoard, ModelCheckpoint

from keras.preprocessing import image
from keras.preprocessing.image import ImageDataGenerator
from keras.applications.imagenet_utils import preprocess_input, decode_predictions

from tensorflow.keras.applications import resnet50

import cv2
import zipfile

import os
import numpy as np

import matplotlib.pyplot as plt

En el zip vienen las imagenes

zip_ref = zipfile.ZipFile('/content/drive/MyDrive/Data_Rural_vs_Urbano.zip', 'r')
zip_ref.extractall('/content/tmp')
zip_ref.close()

train_dir = '/content/tmp/Data_Rural_vs_Urbano/Train'
validation_dir = '/content/tmp/Data_Rural_vs_Urbano/Test'

Definimos los parámetros

width_shape = 224
height_shape = 224
num_classes = 2
epoch = 40
batch_size = 16

Nota: En el ejemplo se utilizan 4 epoch. Si dejas los 40 puede que tarde 4 horas.

Generar imagenes artificiales

train_datagen = ImageDataGenerator(
    rotation_range = 20,
    zoom_range= 0.2,
    width_shift_range = 0.1,
    height_shift_range = 0.1,
    # Rotar la imagen en eje vertical y hotizontal
    horizontal_flip = True,
    vertical_flip = True,

    preprocessing_function = preprocess_input)

valid_datagen = ImageDataGenerator(
    rotation_range = 20,
    zoom_range= 0.2,
    width_shift_range = 0.1,
    height_shift_range = 0.1,
    # Rotar la imagen en eje vertical y hotizontal
    horizontal_flip = True,
    vertical_flip = True,

    preprocessing_function = preprocess_input
)

train_generator = train_datagen.flow_from_directory(
    train_dir,
    target_size = (width_shape, height_shape),
    batch_size = batch_size,
    class_mode = 'categorical'
)
valid_generator = valid_datagen.flow_from_directory(
    validation_dir,
    target_size = (width_shape, height_shape),
    batch_size = batch_size,
    class_mode = 'categorical'
)

Found 878 images belonging to 2 classes.
Found 110 images belonging to 2 classes.

Se guardan los datos:

nb_train_samples = 878
nb_valid_samples = 110

image_input = Input(shape=(width_shape, height_shape, 3)) # Se coloca 3 para los 3 colores

m_Resnet50 = resnet50.ResNet50(input_tensor = image_input, include_top = 'False', weights='imagenet')
m_Resnet50.summary()

last_layer = m_Resnet50.layers[-1].output

x = Flatten(name = 'flatter')(last_layer)
x = Dense(128, activation = 'relu', name = 'fc1')(x)
x = Dropout(0.3)(x)
x = Dense(128, activation = 'relu', name = 'fc2')(x)
x = Dropout(0.3)(x)

out = Dense(num_classes, activation='softmax', name= 'output')(x)
custom_model = Model(image_input, out)
custom_model.summary()