👥 Gender and Age Detection with Deep Learning

Detect age and gender from facial images like a pro! 🚀

Advanced deep learning models for facial attribute analysis with style

---

🎯 What's This?

A powerful deep learning project that analyzes facial images to predict age, gender, and ethnicity. Think of it as your personal facial analysis AI! 🧠

✨ What You Get

• 🎯 Multi-task learning (Age, Gender, Ethnicity)
• 🧠 Deep CNN architecture with TensorFlow/Keras
• 📊 Comprehensive data analysis and visualization
• 🔄 Data augmentation for better model performance
• 📈 Training metrics and model evaluation
• 💾 Model serialization for deployment
• 🎨 Beautiful visualizations with Plotly
• ⚡ Optimized training with callbacks

---

🚀 Quick Start

# 1. Clone it
git clone <your-repo-url>
cd Gender-and-Age-Detection

# 2. Install dependencies
pip install tensorflow pandas numpy matplotlib seaborn plotly scikit-learn

# 3. Run the notebook!
jupyter notebook "Age Detection.ipynb"

That's it! 🎉

---

🎮 How to Use

Option 1: Jupyter Notebook (Recommended)

jupyter notebook "Age Detection.ipynb"

Perfect for interactive development and analysis

Option 2: Google Colab

# Upload the notebook to Google Colab
# Enable GPU runtime for faster training

For cloud-based training with GPU acceleration

Option 3: Command Line Training

# Run specific cells or convert to Python script
python -c "exec(open('Age Detection.ipynb').read())"

For automated training pipelines

---

📊 Sample Output

📊 Dataset Statistics:
- Total samples: 23,705 faces
- Age range: 1-116 years
- Gender distribution: Male/Female balanced
- Ethnicity classes: 5 categories

🧠 Model Performance:
- Gender accuracy: ~52.6%
- Age accuracy: Varies by age group
- Ethnicity accuracy: ~43.2%

📈 Training Metrics:
- Batch size: 64
- Epochs: 100 (with early stopping)
- Learning rate: 0.001 (adaptive)

---

🛠️ What's Inside

Gender-and-Age-Detection/
├── 👥 Age Detection.ipynb          # Main analysis notebook
├── 🧠 age.h5                      # Trained age model
├── 👤 gender.h5                   # Trained gender model
├── 📊 age_gender.csv              # Processed dataset
├── 📚 README.md                   # This file
└── 📄 LICENSE                     # MIT License

---

🎨 Features

📊 Data Analysis & EDA

• Comprehensive dataset exploration
• Age distribution analysis
• Gender and ethnicity balance check
• Missing value detection
• Statistical summaries

🧠 Deep Learning Models

• Multi-task CNN architecture
• Convolutional layers with batch normalization
• Dropout layers for regularization
• Dense layers for classification
• Softmax/Sigmoid activations

🔄 Data Preprocessing

• Image normalization (0-1 scaling)
• Data augmentation (rotation, zoom, shift)
• Train-test splitting (70-30)
• One-hot encoding for categorical variables

📈 Training & Evaluation

• Early stopping to prevent overfitting
• Learning rate reduction on plateau
• Confusion matrix analysis
• Accuracy metrics for each task
• Model serialization (.h5 format)

🎨 Visualizations

• Age distribution histograms
• Gender balance pie charts
• Ethnicity distribution bar plots
• Training curves with Plotly
• Confusion matrices heatmaps

---

🎪 Fun Features

• 🎲 Multi-task learning (3 models in 1)
• 🎮 Interactive visualizations with Plotly
• 🥚 Hidden data insights
• 🎨 Beautiful plotting with seaborn
• 🎯 Real-time training progress
• 🎪 Comprehensive model evaluation

---

🐛 Troubleshooting

Problem: ModuleNotFoundError: No module named 'tensorflow' Solution: pip install tensorflow pandas numpy matplotlib seaborn plotly scikit-learn

Problem: GPU not detected Solution: Install GPU version: pip install tensorflow-gpu

Problem: Memory issues during training Solution: Reduce batch size or use data generators

Problem: Slow training Solution: Enable GPU runtime in Colab or use cloud instances

---

🔧 Technical Highlights

✅ What I Built

• Multi-task CNN for simultaneous prediction
• Data augmentation pipeline for robustness
• Comprehensive EDA with statistical analysis
• Model evaluation with multiple metrics
• Production-ready model serialization
• Scalable architecture for large datasets

🧠 Model Architecture

• Input Layer: 48x48x1 grayscale images
• Conv2D Layers: Feature extraction
• MaxPool2D: Dimensionality reduction
• BatchNormalization: Training stability
• Dropout: Regularization
• Dense Layers: Classification heads

📊 Dataset Features

• Age: 1-116 years (continuous)
• Gender: Male/Female (binary)
• Ethnicity: 5 categories (multi-class)
• Images: 48x48 grayscale pixels

---

📈 Performance Metrics

• Gender Accuracy: ~52.6%
• Age Accuracy: Varies by age group
• Ethnicity Accuracy: ~43.2%
• Training Time: ~30-60 minutes (CPU)
• Model Size: ~10-15 MB per model
• Inference Speed: Real-time capable

---

🤝 Contributing

1. Fork it 🍴
2. Create a branch 🌿
3. Make changes ✏️
4. Submit PR 🚀

Ideas welcome! 💡

---

📊 Data Sources

• Primary Dataset: Age, Gender, and Ethnicity Face Data (CSV)
• Source: Kaggle Dataset by Nipun Arora
• Size: 23,705 facial images
• Format: 48x48 grayscale pixels
• Attributes: Age, Gender, Ethnicity

---

⚠️ Disclaimer

For educational and research purposes! This project analyzes facial images for age, gender, and ethnicity prediction. Always respect privacy and ensure proper consent when using facial analysis technology! 🤖

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

🌟 Star the Repository

If you find this project helpful, please give it a ⭐ on GitHub!

📞 Connect & Support

---

Made with ❤️ and ☕ by Jonathan Thota

Analyzing faces, one pixel at a time! 👥