Cats vs Dogs CNN Classifier

This project implements a Convolutional Neural Network (CNN) to classify images of cats and dogs. It includes data loading, model training, evaluation, and a simple Flask web application for inference.

---

Table of Contents

• Project Overview
• Dataset Structure
• Requirements
• Installation
• Training the Model
• Testing on Single Images
• Running the Web Application
• Project Structure
• Acknowledgements

---

Project Overview

This project builds a CNN to distinguish between cat and dog images. It uses PyTorch for deep learning and torchvision for dataset management. The model is trained on images organized in folders, evaluated on a test split, and saved for inference.

A Flask-based web app allows users to upload images and get real-time predictions.

---

Dataset Structure

Your image dataset should be organized as follows:

data/
└── train/
    ├── cats/
    │   ├── cat1.jpg
    │   ├── cat2.jpg
    │   └── ...
    └── dogs/
        ├── dog1.jpg
        ├── dog2.jpg
        └── ...

• Place all training images in data/train under folders cats and dogs.
• Images should be RGB format and of reasonable size.
• Dataset can be found and organized from Kaggle datasets.

---

Requirements

• Python 3.7+
• PyTorch
• torchvision
• scikit-learn
• Flask (for web app)
• Pillow
• Other common Python libraries

---

Installation

Clone this repository:

git clone https://github.com/PritamTheCoder/Cats_vs_Dogs_CNN.git
cd Cats_vs_Dogs_CNN

Create a virtual environment and activate it:

python -m venv .venv
source .venv/bin/activate      # Linux/Mac
.venv\Scripts\activate         # Windows

Install dependencies:

pip install -r requirements.txt

---

Training the Model

Run the training script:

python train_cnn.py

This will:

• Load and preprocess the dataset from data/train.
• Train the CNN model for 8 epochs (modifiable in the script: train_cnn.py).
• Evaluate accuracy on a held-out test set.
• Save the trained model as cat_v_dog_cnn.pth.

Tips:

• Ensure your dataset directory exists and is correctly structured.
• Use GPU if available for faster training.

---

Testing on Single Images

To test a single image with the trained model:

python test_img.py path/to/your/image.jpg

This script loads the saved model, preprocesses the input image, runs inference, and prints the prediction with confidence.

---

Running the Web Application

The web app allows you to upload an image and get a prediction through a user-friendly interface.

Run the Flask app locally:

python app.py

Open your browser and navigate to:

http://127.0.0.1:5000

Upload an image of a cat or dog, and see the prediction displayed instantly.

---

Hosted Demo

You can also try the live web application hosted at: [https://meoworwoof.onrender.com]

---

Example Predictions

Upload an image of a cat or dog, and the model will classify it instantly. Below are some example screenshots from the web app:

Enrty point of webapp.

Dog image correctly predicted as a dog.

Cat image correctly predicted as a cat.

---

Project Structure

.
├── data/
│   └── train/
│       ├── cats/
│       └── dogs/
│
├── webapp/
│   ├── templates/
│   │   └── index.html           # HTML for app.py Flask web application
│   │
│   ├── static/
│   │   └── style.css            # CSS for app.py Flask web application
│   │
│   ├── cat_v_dog_cnn.pth        # Pretrained model weights (example, for quick testing)
│   ├── app.py                   # Flask web application
│   ├── network.py               # Minimal model architecture just for inference
│
├── train_cnn.py                 # Training script (uses root network.py)
├── test_img.py                   # Single image inference script
├── dataset_download.py           # Script to download & organize dataset from URL
├── dataset_loader.py             # Custom dataset loader (optional)
├── network.py                    # Full model architecture for training
├── requirements.txt              # Virtual environment requirements
├── .gitignore
└── README.md

---

Acknowledgements

• Uses PyTorch and torchvision libraries.
• Inspired by common Cats vs Dogs classification tutorials.
• Thanks to the Dogs vs Cats dataset on Kaggle.