https://github.com/shivamratti13/cat-dog-classification

CNN-Based Cat and Dog Image Classifier
https://github.com/shivamratti13/cat-dog-classification

cnn cnn-classification cnn-keras deep deep-learning image image-classification image-processing python transfer-learning

Last synced: about 2 months ago
JSON representation

CNN-Based Cat and Dog Image Classifier

• Host: GitHub
• URL: https://github.com/shivamratti13/cat-dog-classification
• Owner: shivamratti13
• Created: 2024-11-19T05:49:35.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-11-19T06:03:24.000Z (over 1 year ago)
• Last Synced: 2025-03-23T22:31:47.213Z (over 1 year ago)
• Topics: cnn, cnn-classification, cnn-keras, deep, deep-learning, image, image-classification, image-processing, python, transfer-learning
• Language: Jupyter Notebook
• Homepage:
• Size: 1.83 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# CNN-Based Cat and Dog Image Classifier  

This project demonstrates the development of a binary image classifier to distinguish between cats and dogs using Convolutional Neural Networks (CNN). Leveraging a dataset comprising 20,000 training images and 5,000 validation images, two models were built and evaluated for performance.  

![image](https://github.com/shivamratti13/cat-dog-classification/blob/main/Report/cat%20with%20pretrained.png)

![image](https://github.com/shivamratti13/cat-dog-classification/blob/main/Report/dog%20with%20pretrained.png)

## Project Overview  

1. **Custom CNN Architecture**:  

   - Designed and implemented a CNN model from scratch using the Keras library.  

   - Included convolutional layers, pooling layers, and fully connected layers with ReLU activation.  

   - Employed dropout layers to prevent overfitting.  

   - Achieved a classification accuracy of **77.8%**, showcasing effective model design for a from-scratch approach.

   - ![image](https://github.com/shivamratti13/cat-dog-classification/blob/main/Report/CatvsDog%20without%20dataAugumentation.png)

2. **Pretrained Xception Model**:  

   - Fine-tuned the pretrained Xception model from Keras, leveraging transfer learning.  

   - Replaced the top layers to suit binary classification.  

   - Achieved a remarkable accuracy of **99.05%**, demonstrating the power of transfer learning in image classification.

   - ![image](https://github.com/shivamratti13/cat-dog-classification/blob/main/Report/CatvsDog%20with%20pretrained.png) 

## Key Highlights  

- **Data Preprocessing**: Images were resized, normalized, and augmented with techniques such as rotation and flipping to enhance model generalization.  

- **Evaluation Metrics**: Models were assessed using accuracy, loss values, and confusion matrices.  

- **Comparison of Approaches**: Showcased the performance difference between a custom-built CNN and a state-of-the-art pretrained model.  

This project exemplifies the application of both custom CNN design and transfer learning for efficient image classification. It serves as a foundation for building scalable and high-performing deep learning models.