Malaria Cell Detection Using Deep Learning

Project Overview

This project develops a deep learning model using ResNet architecture to detect malaria-infected cells from microscopic images. The model classifies cell images as either "Parasitized" or "Uninfected" with high accuracy.

Key Features

• Custom ResNet architecture for medical image classification
• Data augmentation to improve model generalization
• Streamlit web application for easy model deployment
• Comprehensive training and evaluation pipeline

Dataset

• Source: Cell Images for Detecting Malaria
• Contains microscopic images of blood cells
• Binary classification: Parasitized vs. Uninfected

Model Architecture

• Residual Network (ResNet) with custom stages
• Key components:
  • Residual modules
  • Batch normalization
  • Dropout for regularization
  • Softmax classification layer

Technical Stack

• Python
• TensorFlow/Keras
• scikit-learn
• OpenCV
• Streamlit

Installation

Prerequisites

• Python 3.8+
• pip

Steps

1. Clone the repository
2. Install dependencies:

pip install -r requirements.txt

3. Download the dataset
4. Run training script:

python train.py

5. Launch Streamlit app:

streamlit run app.py

Model Performance

• Accuracy: 96%
• Precision, Recall: 97%, 98%

Deployment

• Local deployment via Streamlit
• Potential cloud deployment on platforms like Heroku or AWS

Future Work

• Expand dataset
• Experiment with more advanced architectures
• Add multi-class detection capabilities

License

[MIT]