CIFAR-10 – modeling, building, and training

This example shows how to make a CNN for classifying images in the CIFAR-10 dataset. We'll be using a simple convolution neural network implementation of a couple of convolutions and fully connected layers.

Even though the network architecture is very simple, you will see how well it performs when trying to detect objects in the CIFAR-10 images.

So, let's start off this implementation.

Used packages

We import all the required packages for this implementation:

%matplotlib inline
%config InlineBackend.figure_format = 'retina'

from urllib.request import urlretrieve
from os.path import isfile, isdir
from tqdm import tqdm
import tarfile
import numpy as np
import random
import matplotlib.pyplot as plt
from sklearn.preprocessing import LabelBinarizer
from sklearn.preprocessing import OneHotEncoder

import pickle
import tensorflow as tf

Loading the CIFAR-10 dataset

In this implementation, we'll use CIFAR-10, which is one of the most widely used datasets for...