Hackers Realm
Dogs vs Cats Image Classification using Python | CNN | Deep Learning Project Tutorial
Updated: 1 day ago
Embark on the journey of image classification with Python! This tutorial explores CNN and deep learning techniques to classify images of dogs and cats. Learn to build accurate models that can distinguish between these furry friends, unlocking applications in pet recognition, animal monitoring, and more. Enhance your skills in computer vision, deep learning, and unleash the power of image classification. Join this comprehensive project tutorial to unravel the world of dogs vs cats image classification. #DogsVsCats #Python #CNN #DeepLearning #ImageClassification #ComputerVision

In this project tutorial we will use Convolutional Neural Network (CNN) for image feature extraction and visualize the results with plot graphs.
You can watch the video-based tutorial with step by step explanation down below.
Dataset Information
The training archive contains 25,000 images of dogs and cats. Train your algorithm on these files and predict the labels
(1 = dog, 0 = cat).
Download the dataset here
Environment: Google Colab
Download Dataset
We can download the dataset directly from the Microsoft page
!wget https://download.microsoft.com/download/3/E/1/3E1C3F21-ECDB-4869-8368-6DEBA77B919F/kagglecatsanddogs_3367a.zipa
--2021-05-06 16:04:20-- https://download.microsoft.com/download/3/E/1/3E1C3F21-ECDB-4869-8368-6DEBA77B919F/kagglecatsanddogs_3367a.zip Resolving download.microsoft.com (download.microsoft.com)... 23.78.216.154, 2600:1417:8000:980::e59, 2600:1417:8000:9b2::e59 Connecting to download.microsoft.com (download.microsoft.com)|23.78.216.154|:443... connected. HTTP request sent, awaiting response... 200 OK Length: 824894548 (787M) [application/octet-stream] Saving to: ‘kagglecatsanddogs_3367a.zip’ kagglecatsanddogs_3 100%[===================>] 786.68M 187MB/s in 4.3s 2021-05-06 16:04:24 (183 MB/s) - ‘kagglecatsanddogs_3367a.zip’ saved [824894548/824894548]
Unzip the Dataset
!unzip kagglecatsanddogs_3367a.zip
Run this code once and comment it
Import Modules
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import warnings
import os
import tqdm
import random
from keras.preprocessing.image import load_img
warnings.filterwarnings('ignore')
pandas - used to perform data manipulation and analysis
numpy - used to perform a wide variety of mathematical operations on arrays
matplotlib - used for data visualization and graphical plotting
os - used to handle files using system commands
tqdm - progress bar decorator for iterators
random - used for randomizing
load_img - used for loading the image as numpy array
warnings - to manipulate warnings details, filterwarnings('ignore') is to ignore the warnings thrown by the modules (gives clean results)
Create Dataframe for Input and Output
The Dogs vs Cats dataset may differ from where it was downloaded like folder structures or labels. You may create a dataframe to convert the input and output paths accordingly for easier processing.
input_path = []
label = []
for class_name in os.listdir("PetImages"):
for path in os.listdir("PetImages/"+class_name):
if class_name == 'Cat':
label.append(0)
else:
label.append(1)
input_path.append(os.path.join("PetImages", class_name, path))
print(input_path[0], label[0])
PetImages/Dog/4253.jpg 1
Adding the label to the images, one (1) for dogs and zero (0) for cats
Display the path of first image with corresponding label
Now we create the dataframe for processing
df = pd.DataFrame()
df['images'] = input_path
df['label'] = label
df = df.sample(frac=1).reset_index(drop=True)
df.head()

Display of image paths with labels
Data was shuffled and the index was removed
We must remove any files in the data set that are not image data to avoid errors
for i in df['images']:
if '.jpg' not in i:
print(i)
PetImages/Cat/Thumbs.db PetImages/Dog/Thumbs.db
import PIL
l = []
for image in df['images']:
try:
img = PIL.Image.open(image)
except:
l.append(image)
l
['PetImages/Cat/666.jpg', 'PetImages/Cat/Thumbs.db', 'PetImages/Dog/Thumbs.db', 'PetImages/Dog/11702.jpg']
List of non-image type files and corrupted images
# delete db files
df = df[df['images']!='PetImages/Dog/Thumbs.db']
df = df[df['images']!='PetImages/Cat/Thumbs.db']
df = df[df['images']!='PetImages/Cat/666.jpg']
df = df[df['images']!='PetImages/Dog/11702.jpg']
len(df)
24998
Dropping the corrupted files and non-image files from the dataset
Exploratory Data Analysis
Let us display a grid of images to know the content of the data
# to display grid of images
plt.figure(figsize=(25,25))
temp = df[df['label']==1]['images']
start = random.randint(0, len(temp))
files = temp[start:start+25]
for index, file in enumerate(files):
plt.subplot(5,5, index+1)
img = load_img(file)
img = np.array(img)
plt.imshow(img)
plt.title('Dogs')
plt.axis('off')





Display of 25 random images of dogs
plt.axis('off') turns off both axis from the images
Files loaded and stored in an array
# to display grid of images
plt.figure(figsize=(25,25))
temp = df[df['label']==0]['images']
start = random.randint(0, len(temp))
files = temp[start:start+25]
for index, file in enumerate(files):
plt.subplot(5,5, index+1)
img = load_img(file)
img = np.array(img)
plt.imshow(img)
plt.title('Cats')
plt.axis('off')





Display of 25 random images of cats
Different saturation and qualities among the images
import seaborn as sns
sns.countplot(df['label'])

seaborn - built on top of matplotlib with similar functionalities
We can observe an equal distribution of both classes
Create Data Generator for the Images
Data Generators loads the data from the disk for reading and training the data directly, saving RAM space and avoiding possible overflow that might crash the system.
df['label'] = df['label'].astype('str')
df.head()

Convert the data type of 'label' to string for easier processing
Let us split the input data
# input split
from sklearn.model_selection import train_test_split
train, test = train_test_split(df, test_size=0.2, random_state=42)
from keras.preprocessing.image import ImageDataGenerator
train_generator = ImageDataGenerator(
rescale = 1./255, # normalization