README.md

# AI Malaria Detection System - Ghana Model

## Overview
This system uses deep learning (MobileNetV2) to detect malaria parasites in blood smear images. The model has been fine-tuned specifically on Ghanaian blood smear samples to ensure accuracy with local staining and imaging conditions.

## Performance Metrics
- **Overall Accuracy**: 97.2%
- **Parasitized Detection**: 97.8%
- **Uninfected Rejection**: 92.3%
- **Model Size**: 2.54 MB (mobile-optimized)

## Folder Structure
```
/malaria_detector/
  ├── models/
  │   ├── local_finetuned_model.h5
  │   └── malaria_model.tflite
  ├── scripts/
  │   ├── predict_patches.py
  │   └── requirements.txt
  ├── evaluation/
  │   ├── final_report.pdf
  │   └── confusion_matrix.png
  └── README.md
```

## Deployment & Usage
1.  **Clone the Repository**: Download the entire `malaria_detector` folder.
2.  **Install Dependencies**: Navigate to the `scripts/` folder and install required libraries:
    ```bash
    pip install -r requirements.txt
    ```
3.  **Model Location**: Ensure `local_finetuned_model.h5` and `malaria_model.tflite` are in the `models/` folder.
4.  **Run Prediction**: Use the `predict_patches.py` script to analyze whole slide images:
    ```bash
    python scripts/predict_patches.py /path/to/your/image.jpg [confidence_threshold]
    ```
    *   `[confidence_threshold]` is optional; defaults to `0.5` (50%). Recommended: `0.5`.

## Contact
For questions or further assistance, please contact the development team.