🩸 Blood Cell Classification & GAN Image Generation

A professional dark-themed web application to classify blood cell images using ViT, Custom ViT, and Performer models, with GAN-based image generation. Fully tracked using MLflow.

---

🌟 Features

• Image Classification
  • Predict blood cell types using:
    • Pretrained ViT
    • Custom ViT
    • Performer (Efficient Transformer)
• GAN Image Generation
  • Generate synthetic blood cell images for data augmentation or visualization
• MLflow Integration
  • Track experiments, metrics, and save models
• Professional Frontend
  • Dark-themed UI
  • Real-time predictions and GAN generation

---

🖼️ Demo Screenshots

Upload & Predict:

GAN Image Generation:

Add your screenshots to screenshots/ folder.

---

📂 Folder Structure

blood-cell-classification/
│
├─ app.py                  # Flask API for predictions and GAN
├─ templates/
│   └─ index.html          # Frontend
├─ static/
│   └─ gan_image.png       # Generated GAN image
├─ models/
│   ├─ vit_model.pth
│   ├─ custom_vit_model.pth
│   └─ performer_model.pth
├─ your_model_file.py      # All model classes
├─ notebooks/
│   └─ training_notebooks.ipynb
├─ mlruns/                 # MLflow experiments
├─ requirements.txt
└─ .gitignore

---

⚙️ Installation

1. Clone the repository:

git clone https://github.com/ankitsunil530/blood-cell-classification.git
cd blood-cell-classification

2. Create a virtual environment:

python -m venv venv
source venv/bin/activate  # Linux/Mac
venv\Scripts\activate     # Windows

3. Install dependencies:

pip install -r requirements.txt

---

🚀 Running the Application

python app.py

Open in browser:

http://127.0.0.1:5000/

Features:

• Upload blood cell images → Predict classes with three models
• Generate GAN images → Real-time synthetic image generation

---

🧠 Models

Model	Description
BloodCellViT	Pretrained ViT-Base/16 fine-tuned on dataset
CustomViT	ViT trained from scratch for 4 blood cell classes
Performer	Efficient transformer with linear attention
GAN	DCGAN-based generator for synthetic blood cells

---

📊 Training & MLflow

• Training notebooks in notebooks/
• MLflow logs in mlruns/
• Log model & metrics:

mlflow.pytorch.log_model(model, artifact_path="model_name")
mlflow.log_param("learning_rate", 0.0001)
mlflow.log_metric("val_accuracy", 97.12)

---

🎨 Frontend

• Dark theme with professional look

• Displays:

  • Uploaded image predictions
  • GAN-generated images

• Real-time updates

---

📌 Notes

• Place trained model files (.pth) in models/ folder
• MLflow tracking URI:

mlflow.set_tracking_uri("file:///D:/Blood Cell Classifiaction/blood-cell-classification/mlruns")

• GAN images saved in static/gan_image.png

---

🔮 Future Improvements

• Batch predictions for multiple images
• Higher-resolution GAN generation
• Docker deployment for cloud hosting
• User authentication and multi-user support

---

📧 Contact

Sunil Kumar

• GitHub: ankitsunil530
• Email: sunilkumar@example.com

---

📝 License

This project is licensed under the MIT License.

---