Skip to content

README.md

Latest commit

 

History

History
34 lines (27 loc) · 1.83 KB

README.md

File metadata and controls

34 lines (27 loc) · 1.83 KB
Raw

🖼️ Image Classification using CNN (CIFAR-10)

📌 Project Overview

In this project, I developed a Convolutional Neural Network (CNN) to classify low-resolution (32x32) color images into 10 distinct categories. This project demonstrates the power of Deep Learning in identifying spatial hierarchies and patterns in visual data.

🛠️ Tech Stack

  • Framework: TensorFlow / Keras
  • Libraries: NumPy, Matplotlib, Seaborn, Scikit-Learn
  • Dataset: CIFAR-10 (60,000 images across 10 classes)

🏗️ Model Architecture

The network consists of a series of convolutional and pooling layers designed to extract features from edges to complex shapes:

  1. Convolutional Layer 1: 32 filters (3x3), ReLU activation.
  2. Max Pooling: 2x2 downsampling.
  3. Convolutional Layer 2: 64 filters (3x3), ReLU activation.
  4. Max Pooling: 2x2 downsampling.
  5. Convolutional Layer 3: 64 filters (3x3).
  6. Fully Connected (Dense): 64 neurons leading to a 10-node output layer.

📊 Performance & Evaluation

  • Final Test Accuracy: 69.93% after 10 epochs.
  • Optimization: Used Adam optimizer and SparseCategoricalCrossentropy loss.

Confusion Matrix Analysis

Confusion Matrix

By analyzing the confusion matrix, I identified key behaviors of the model:

  • High Precision: The model performed exceptionally well on Ships (835 correct) and Automobiles (790 correct).
  • Class Confusion: The highest error rate occurred between Cats and Dogs, where 264 cats were misidentified as dogs, highlighting the challenges of classifying similar organic shapes in low resolution.

🚀 How to Run

  1. Clone the repository.
  2. Install dependencies: pip install tensorflow matplotlib seaborn scikit-learn.
  3. Run the notebook: jupyter notebook image-classification-cnn.ipynb.