Gender Detection on masked faces using CNN

📌 Project Overview

This project implements a Convolutional Neural Network (CNN) for gender classification based on facial images. The model is trained on a dataset containing images of individuals wearing masks. The goal is to predict whether the person is Male or Female using deep learning techniques.

📂 Project Structure

├── Dataset/                # Folder containing images
│   ├── Female/             # Images of female individuals
│   ├── Male/               # Images of male individuals
├── CNN_Model.ipynb         # Jupyter Notebook containing training and evaluation
├── plot_accuracy_loss.png  # plot of accuracy and loss
├── masked_sample           # sample of images used for training
├── requirements.txt        # Dependencies required to run the project
└── README.md               # Project documentation

🚀 Getting Started

1️⃣ Clone the Repository

git clone https://github.com/MohaYass92/Gender-Detection-on-Masked-Faces-with-CNN
cd Gender-Detection-on-Masked-Faces-with-CNN

2️⃣ Install Dependencies

Ensure you have Python 3.x installed, then run:

pip install -r requirements.txt

3️⃣ Open and Run the Notebook

You can run the Jupyter Notebook in Google Colab or locally:

jupyter notebook

Then open CNN_Model.ipynb and execute the cell.

🏋️ Training the Model

The notebook contain only one cell and it includes:

• Load and preprocess the dataset
• Define and train a CNN model using TensorFlow/Keras
• Evaluate the model on the dataset
• Save the trained model for later use

📦 Dataset

The dataset contains images of masked individuals, divided into two folders:

• Female/
• Male/

Example of Masked Image

Ensure that your dataset is correctly placed before running the notebook.

💾 Saving and Loading the Model

The trained model is saved in .keras format. You can load and use it later:

from tensorflow.keras.models import load_model
model = load_model('CNN_Model.keras')

📊 Model Performance

The training accuracy and loss curves are shown below:

📜 License

This project is open-source. Feel free to modify and improve it!