UTKFace Age, Gender, and Race Classification

Overview

This project trains a deep learning model to predict age, gender, and race from facial image.

Dataset UTKFace Dataset on Kaggle

Samples of the data

• 

Facial images are labeled with:

• Age: 0–116
• Gender: 0 (Male), 1 (Female)
• Race: 0 (White), 1 (Black), 2 (Asian), 3 (Indian), 4 (Others)

EDA Example Age Distribution

• 

Workflow

1. Preprocessing

• Extracts age, gender, and race from filenames
• Filters unwanted race categories and saves the cleaned dataset to Images.csv

2. Image Processing

• Resizes images to 168x168 pixels
• Normalizes pixel values to the [0,1] range

3. Model Training

• Uses a CNN built with TensorFlow/Keras
• Includes Conv2D, MaxPooling, Dense, and Dropout layers
• Optimized using Adam and mean squared error (MSE) loss

4. Evaluation & Inference

• Evaluates model performance on a test split
• Predicts age on new images (e.g., Sharif.png)

Requirements

pip install numpy pandas matplotlib tensorflow pillow scikit-learn opencv-python

Running the Project

1. Place the UTKFace dataset in the working directory
2. Run the script to preprocess, train, and evaluate
3. Use the trained model for inference

Output

• Images.csv with processed data
• Trained model with evaluation metrics
• Data distribution and prediction insights throughout this README

Author

Sharif