Cats and Dogs Image Classification using SVM and CNN Models

July 1, 2023

Overview

As part of a research paper on "Comparative Analysis of SVM and CNN in Image Classification," I developed and evaluated models for classifying images of cats and dogs. This project involved the use of both Support Vector Machine (SVM) and Convolutional Neural Network (CNN) models.

Project Components

Models

  • Support Vector Machine (SVM):

    • File: SVM4000.ipynb
    • Description: This notebook details the implementation and evaluation of the SVM model trained on 4,000 images of cats and dogs.

  • Convolutional Neural Network (CNN):

    • File: CNN4000.ipynb
    • Description: This notebook demonstrates the CNN model trained and tested on a subset of 4,000 images.
    • File: CNN20000.ipynb
    • Description: This notebook showcases the CNN model trained and tested on a larger dataset of 20,000 images.

Accuracy and Loss

The accuracy and loss metrics for the models are as follows:

| Model | Accuracy | Loss | |--------------|----------|-------| | CNN (20,000) | 89% | 26% | | CNN (4,000) | 76% | 50% | | SVM (4,000) | 61% | - |

Technologies

  • Programming Language: Python
  • Libraries and Frameworks:
    • TensorFlow
    • Keras
    • Scikit-learn
    • Matplotlib
    • NumPy
    • Pandas
  • Tools:
    • Jupyter Notebook
    • Git
    • GitHub

Publication

This research will be published in the upcoming conference, IEEE NMITCON-2023.

Repository

You can find the complete code and resources for this project in the GitHub repository.

Created as part of a research paper on "Comparative Analysis of SVM and CNN in Image Classification" to be published in IEEE NMITCON-2023.