🔬 White Blood Cell Classification using CNN

A deep learning project implementing a Convolutional Neural Network (CNN) for automated classification of white blood cells from microscopic images, achieving 95% accuracy across five cell types.

Author: Muhammad Husnain Ali

🛠️ Technologies Used

Core Technologies

• Python - Primary programming language
• TensorFlow - Deep learning framework
• Keras - Neural network API
• Google Colab - Development environment

Data Processing & Model

• NumPy - Numerical computing
• OpenCV - Image processing
• Matplotlib - Visualization
• Scikit-learn - Model evaluation

🚀 Features

• Advanced Image Processing

  • Automatic image resizing (128x128)
  • Pixel normalization
  • Data augmentation with rotations and flips
  • Balanced class distribution

• Robust CNN Architecture

  • 4-stage convolutional network
  • Batch normalization
  • Dropout regularization
  • MaxPooling for feature selection

• Performance Optimization

  • Early stopping
  • Learning rate scheduling
  • Model checkpointing
  • GPU acceleration

• Comprehensive Evaluation

  • Confusion matrix analysis
  • ROC curve generation
  • Per-class accuracy metrics
  • Real-time prediction visualization

🏗️ Model Architecture

Input (128x128x3)
│
├── Conv Block 1 (64 filters) → BatchNorm → MaxPool → Dropout(0.25)
│
├── Conv Block 2 (128 filters) → BatchNorm → MaxPool → Dropout(0.25)
│
├── Conv Block 3 (256 filters) → BatchNorm → MaxPool → Dropout(0.25)
│
├── Conv Block 4 (512 filters) → BatchNorm → MaxPool → Dropout(0.25)
│
├── Dense(1024) → Dropout(0.5)
│
├── Dense(512) → Dropout(0.5)
│
└── Output(5) → Softmax

📊 Dataset

• Source: White Blood Cells Dataset
• Classes: 5 WBC types
  • Neutrophils
  • Eosinophils
  • Basophils
  • Monocytes
  • Lymphocytes
• Volume: 14,514 images
  • Training: 8,140 (80%)
  • Validation: 2,035 (20%)
  • Test: 4,339

📈 Performance

Metric	Score
Training Accuracy	93.73%
Validation Accuracy	94.94%
Test Accuracy	95.00%
Inference Time	<100ms

🚀 Quick Start

1. Open in Colab

2. Mount Google Drive

from google.colab import drive
drive.mount('/content/drive')

3. Install Dependencies

!pip install -r requirements.txt

4. Download Dataset

!kaggle datasets download -d paultimothymooney/blood-cells
!unzip blood-cells.zip

5. Run Training

!python train.py --epochs 10 --batch_size 32

📦 Requirements

tensorflow>=2.8.0
numpy>=1.19.5
opencv-python>=4.5.5
matplotlib>=3.4.3
scikit-learn>=1.0.2

💡 Usage Tips

Training

• Use GPU runtime in Colab
• Enable data augmentation
• Start with provided hyperparameters
• Monitor validation metrics

Prediction

model = load_model('wbc_classifier.h5')
prediction = model.predict(image)

Evaluation

from evaluation import evaluate_model
metrics = evaluate_model(model, test_data)

🤝 Contributing

1. Fork the repository
2. Create feature branch (git checkout -b feature/AmazingFeature)
3. Commit changes (git commit -m 'Add AmazingFeature')
4. Push to branch (git push origin feature/AmazingFeature)
5. Open Pull Request

📞 Support

For questions and support:

• Open an Issue
• Contact: m.husnainali.work@gmail.com

🙏 Acknowledgments

• Kaggle for the comprehensive dataset
• TensorFlow team for the framework
• Google Colab for free GPU resources
• All contributors to this project

---

Made with ❤️ by Muhammad Husnain Ali