🤖 ML Image Classifier - Handwritten Digit Recognition

📋 Overview

Welcome to ML Image Classifier! 🎉 This beginner-friendly machine learning project demonstrates how to build a handwritten digit classifier using Support Vector Machine (SVM) algorithm. The model achieves >95% accuracy on the scikit-learn digits dataset, classifying digits from 0 to 9.

Perfect for:

• 🎓 Machine Learning beginners
• 📚 Students learning image classification
• 🔬 Researchers exploring SVM algorithms
• 💡 Anyone interested in computer vision

✨ Features

• 🚀 Quick Setup: Run in minutes with minimal dependencies
• 📊 High Accuracy: Achieves >95% accuracy on test data
• 📈 Visualization: Beautiful plots showing predictions and confusion matrix
• 🎯 Clean Code: Well-commented and easy to understand
• 📱 Lightweight: Uses only 8x8 pixel images (64 features)

🖼️ Sample Output

The classifier correctly identifies handwritten digits with high accuracy. Green titles indicate correct predictions, while red indicates misclassifications.

🛠️ Tech Stack

Python

Scikit-learn

Matplotlib

NumPy

🚀 Quick Start

Prerequisites

• Python 3.8 or higher
• pip package manager

Installation

1. Clone the repository

   git clone https://github.com/ShamsRupak/ML-Image-Classifier.git
   cd ML-Image-Classifier

2. Install dependencies

   pip install -r requirements.txt

3. Run the classifier

   python classifier.py

That's it! 🎉 The script will:

• Load the digits dataset
• Train an SVM model
• Display accuracy metrics
• Show sample predictions with matplotlib

📊 How It Works

Click to expand the algorithm explanation

1. Data Loading 📥

The script loads the built-in digits dataset from scikit-learn, containing 1,797 samples of 8×8 grayscale images.

2. Data Preprocessing 🔧

• Images are flattened from 8×8 matrices to 64-dimensional vectors
• Data is split into 80% training and 20% testing sets

3. Model Training 🤖

• Uses Support Vector Machine (SVM) with RBF kernel
• Hyperparameters: gamma=0.001, kernel='rbf'
• Trains on 1,437 samples

4. Evaluation 📈

• Makes predictions on 360 test samples
• Calculates accuracy and generates classification report
• Creates visualizations

📈 Performance Metrics

The model typically achieves:

• Overall Accuracy: ~97-99%
• Per-digit Precision: >95% for most digits
• Training Time: <1 second on modern hardware

Sample Classification Report

              precision    recall  f1-score   support

           0       1.00      1.00      1.00        36
           1       0.97      0.97      0.97        36
           2       1.00      1.00      1.00        36
           3       0.97      0.92      0.94        36
           4       0.97      1.00      0.99        36
           5       0.95      0.97      0.96        36
           6       1.00      1.00      1.00        36
           7       0.97      0.97      0.97        36
           8       0.94      0.94      0.94        36
           9       0.94      0.94      0.94        36

    accuracy                           0.97       360
   macro avg       0.97      0.97      0.97       360
weighted avg       0.97      0.97      0.97       360

🎯 Use Cases

• Educational Projects: Learn ML fundamentals
• Digit Recognition Apps: Foundation for OCR systems
• Research: Baseline for comparing algorithms
• Hackathons: Quick ML demo project

🤝 Contributing

Contributions are welcome! Feel free to:

• 🐛 Report bugs
• 💡 Suggest new features
• 🔧 Submit pull requests

📚 Learning Resources

Want to learn more? Check out these resources:

• Scikit-learn Documentation
• SVM Explained
• Machine Learning Course by Andrew Ng

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

👤 Author

Shams Rupak

• GitHub: @ShamsRupak
• Project Link: https://github.com/ShamsRupak/ML-Image-Classifier

---

Made with ❤️ by Shams Rupak | ⭐ Star this repo if you found it helpful!