Malar.ai - Malaria Detection Using Modified ResNet50

This project aims to detect malaria from cell images using a modified version of the ResNet50 architecture to process grayscale images.

Overview

Malaria is a life-threatening disease caused by parasites transmitted to humans through the bites of infected female Anopheles mosquitoes. Early and accurate detection is crucial for effective disease management.

In this project, I adapt the ResNet50 architecture, pre-trained on ImageNet, to process grayscale images for malaria detection.

Features

• Data Processing: Images are loaded, resized to 224x224 pixels, and converted to grayscale.
• Customized ResNet50: I have modified the ResNet50 architecture to accept single-channel (grayscale) images. The single channel is then expanded to mimic a 3-channel image to utilize the ResNet50 architecture more effectively.
• Normalization: Data is normalized based on the average of the ImageNet RGB means to align with the expectations of the pre-trained weights.

Prerequisites

• Python 3.x
• OpenCV (cv2)
• Keras
• TensorFlow
• scikit-learn
• Cell Dataset, which can be found here

Usage

1. Clone the repository:

   git clone https://github.com/ArmyyA/Malar.ai.git
   cd Malar.ai

2. Place your dataset in the appropriate directory (images/Parasitized for positive samples and images/Uninfected for negative samples).

3. Run the script:

   python main.py

4. The trained model will be saved as malaria_detection_model.h5.

Metrics and Results

The model performance was evaluated using the following metrics:

• Loss: Binary Crossentropy
• Accuracy: Percentage of correctly classified images

Training Results:

• Prediction Accuracy: 97%