Advanced Blood Cell Classification with AI

HematoVision is an innovative AI-powered web application designed for the precise and efficient classification of blood cells. This project aims to assist pathologists and healthcare professionals in rapid diagnostics and research by automatically identifying different types of blood cells from uploaded images.

Developed as part of the Smart Bridge Internship Program.

Key Features

Core functionalities that make HematoVision powerful and efficient.

Accurate Classification

Utilizes a fine-tuned CNN for high accuracy in identifying various blood cell types.

Transfer Learning

Employs the MobileNetV2 architecture to ensure efficient training and robust performance.

Intuitive Web Interface

A user-friendly web app built with Flask for easy image uploads and result viewing.

Real-time Feedback

Provides immediate predictions, including class and confidence score, upon submission.

Scalable Approach

The underlying deep learning model is designed to be efficient for practical application.

Open Source Stack

Built entirely on powerful and widely-used open source technologies.

Technology & Data

This project is built on a foundation of robust, open-source technologies. The model was trained on a balanced dataset of blood cell images, which is crucial for achieving unbiased classification performance.

Python TensorFlow Keras Flask NumPy Pandas OpenCV

Training Dataset Composition

How It Works

From setup to execution, here's a look under the hood.

Setup Instructions

1. Prerequisites: Ensure you have Anaconda Navigator and Git installed.

2. Create Conda Environment:

conda create -n hemato_env python=3.9\nconda activate hemato_env

3. Install Dependencies:

Create a `requirements.txt` file and run the following command:

pip install -r requirements.txt

Project Directory Structure

HematoVision/
├── app.py              # Flask web application
├── trained_model.h5    # The trained ML model
├── requirements.txt    # Python dependencies
├── static/             # For uploaded images
├── templates/
│   ├── index.html
│   └── result.html
└── README.md           # Project documentation

Running the Application

1. Navigate to Directory:

cd C:\HematoVision

2. Activate Environment:

conda activate hemato_env

3. Start the Flask Server:

python app.py

4. Access in Browser: Open a web browser and go to http://127.0.0.1:5000.

Future Vision

Potential enhancements to expand the project's capabilities.

➔

Batch Prediction: Allow users to upload multiple images for simultaneous classification.

➔

Detailed Reports: Generate downloadable PDF reports with classification results and probabilities.

➔

Containerization: Package the application using Docker for easy, consistent deployment.

➔

Cloud Deployment: Host the application on a public cloud platform like Heroku or AWS.