Blood Cell Diagnostics

End-to-End AI platform for classifying blood cell images (+95% accuracy): 17K+ images (8 classes), custom PyTorch CNN, Flask web app, Docker deploy.

Overview

Blood Cell Diagnostics is a comprehensive AI-powered medical imaging platform designed to classify blood cell images with high accuracy. The system processes 17,000+ images across 8 different blood cell classes, enabling rapid and accurate diagnosis. Built with a custom PyTorch CNN architecture, the platform features a Flask web application with Docker deployment for seamless production use.

Challenge

Manual blood smear review is slow and depends on specialist time. The goal was to classify microscopy images across 8 blood cell types reliably enough to support faster screening, and to make the model usable through a simple web interface rather than a notebook.

Solution

Custom Convolutional Neural Network (CNN) trained on 17,000+ blood cell images across 8 classes, delivered as a Flask web application with Docker deployment.

Architecture

Custom PyTorch CNN · OpenCV / NumPy / PIL preprocessing · Flask inference API · Docker production deploy

Results

  • Accuracy: 95%+
  • Classes: 8
  • Dataset Size: 17K+ images
  • Achieved 95%+ accuracy on blood cell classification
  • Custom CNN architecture optimized for medical imaging
  • Production-ready Flask web application
  • Docker containerization for easy deployment
  • Real-time image processing capabilities

Technologies

  • Python
  • PyTorch
  • OpenCV
  • Flask
  • Docker
  • HTML/CSS/JS

All projects · Contact

Loading project…