Cat-Dog-Image-Classifier-CNN

Convolutional Neural Network (CNN) built with TensorFlow/Keras to classify cat and dog images, achieving 68.0% accuracy.

Cat and Dog Image Classifier (Convolutional Neural Network)

Project Overview

This project involves building, training, and optimizing a Convolutional Neural Network (CNN) to correctly classify images of cats and dogs. This challenge was completed as part of the freeCodeCamp Machine Learning with Python Certification.

• Final Accuracy: 68.0% (Passed the required 63% threshold).
• Methodology: Used Transfer Learning principles and Keras for rapid model prototyping and training.

---

Technical Stack

• Primary Language: Python
• Frameworks: TensorFlow 2.x, Keras
• Libraries: NumPy, Matplotlib
• Environment: Google Colaboratory (GPU Acceleration)

---

Model Architecture & Techniques

The final model was optimized to prevent overfitting on the small dataset through aggressive regularization and architectural deepening.

Key Components:

• Deep CNN Structure: A 4-block stack of Conv2D and MaxPooling2D layers (with filters up to 256) was used to ensure the model had enough capacity to extract complex features.
• Data Augmentation: The training dataset was artificially expanded using random image transformations (rotation, zoom, flips) to increase diversity and prevent memorization.
• Aggressive Regularization: Dropout layers (set to 0.4) were placed after every MaxPooling2D layer in the feature extraction stack, forcing the model to generalize and avoid overreliance on specific patterns.

---

🔗 Live Code & Execution

Click the link below to view the executable Google Colab notebook.