🪖 Helmet Compliance Monitoring System

Automated Safety Compliance Through Computer Vision

Safety rule violations, like not wearing helmets, are hard to track manually. A computer vision system can automate helmet compliance monitoring, enhance workplace safety, and reduce accident severity without needing extra staff.

🎯 Problem Statement

Traditional manual safety monitoring faces several challenges:

• Human Error: Manual oversight can miss violations
• Resource Intensive: Requires dedicated safety personnel
• Inconsistent Monitoring: Cannot provide 24/7 surveillance
• Delayed Response: Violations detected after incidents occur
• Documentation Issues: Difficult to maintain compliance records

💡 Solution

Our AI-powered system provides:

• Real-time Detection: Instant helmet compliance verification
• Automated Monitoring: Continuous surveillance without human intervention
• Accurate Documentation: Automated violation logging and reporting
• Cost-effective: Reduces need for additional safety staff
• Scalable: Can monitor multiple locations simultaneously

🚀 Key Features

🔍 Detection Capabilities

• Real-time Helmet Detection - Instant compliance verification
• Live Camera Integration - Continuous monitoring through webcam/IP cameras
• Batch Image Processing - Analyze multiple images simultaneously
• High Accuracy - Advanced AI model with adjustable confidence thresholds

📊 Monitoring & Analytics

• Violation Logging - Automatic incident recording with timestamps
• Compliance Reporting - Generate detailed safety compliance reports
• Statistics Dashboard - Track compliance rates and trends
• Export Functionality - Download reports in CSV format

⚙️ User-Friendly Interface

• Intuitive Web Interface - Easy-to-use Streamlit dashboard
• Adjustable Thresholds - Customize detection sensitivity
• Multiple Input Methods - Upload images, use camera, or batch process
• Real-time Feedback - Instant safety status notifications

🏭 Use Cases

Industrial Applications

• Construction Sites - Monitor workers in hard hat zones
• Manufacturing Plants - Ensure safety compliance in production areas
• Warehouses - Automated safety checks in material handling zones
• Mining Operations - Critical safety monitoring in hazardous environments

Benefits for Organizations

• Reduced Accidents - Proactive safety violation prevention
• Lower Insurance Costs - Improved safety records
• Regulatory Compliance - Meet OSHA and safety standards
• Enhanced Productivity - Automated monitoring frees staff for other tasks

📋 Technical Requirements

System Requirements

• Python: 3.8 or higher
• RAM: Minimum 4GB (8GB recommended)
• Storage: 500MB free space
• Camera: Optional (for live detection)

Dependencies

streamlit==1.28.0
tensorflow==2.13.0
opencv-python==4.8.0.74
Pillow==10.0.0
numpy==1.24.3
pandas==2.0.3

🔧 Installation Guide

1. Clone Repository

git clone https://github.com/HarshitBhalani/Helmet-Compliance-Monitoring.git
cd Helmet-Compliance-Monitoring

2. Create Virtual Environment (Recommended)

# Windows
python -m venv helmet_env
helmet_env\Scripts\activate

# macOS/Linux
python -m venv helmet_env
source helmet_env/bin/activate

3. Install Dependencies

pip install -r requirements.txt

4. Model Setup

⚠️ Important: Model files are not included in this repository due to size constraints.

Option A: Train Your Own Model

1. Visit Google Teachable Machine
2. Create image classification project
3. Upload helmet/no-helmet training images
4. Train and export as TensorFlow model
5. Download and rename to model.h5
6. Place in project root directory

Option B: Use Pre-trained Model (If Available)

1. Contact repository maintainer for model file
2. Place model.h5 in project root directory

5. Run Application

streamlit run app.py

6. Access Interface

Open your browser and navigate to: http://localhost:8501

📱 How to Use

1. Single Image Detection

• Select "📷 Image Upload" mode
• Upload an image file (JPG, PNG, BMP)
• View detection results and confidence scores
• System automatically logs violations

2. Live Camera Monitoring

• Select "📹 Live Camera" mode
• Allow camera permissions
• Capture photos for real-time analysis
• Get instant safety compliance feedback

3. Batch Processing

• Select "📁 Batch Processing" mode
• Upload multiple images at once
• Generate comprehensive compliance report
• Export results for documentation

4. View Reports

• Select "📊 Violation Logs" mode
• Review all detected violations
• Analyze compliance trends
• Export data for regulatory reporting

🎛️ Configuration Options

Detection Threshold Settings

• 0.5-0.6: Lenient (reduces false violations, may miss some cases)
• 0.7: Balanced (recommended for most environments)
• 0.8-0.9: Strict (ideal for high-risk environments)

Customization Tips

• Adjust threshold based on your safety requirements
• Higher thresholds for critical safety zones
• Lower thresholds for general monitoring areas

🤖 AI Model Information

Model Specifications

• Architecture: Convolutional Neural Network (CNN)
• Input Size: 224x224 pixels
• Classes: 2 (With Helmet, Without Helmet)
• Framework: TensorFlow/Keras
• Training: Supervised learning on labeled helmet images

Performance Metrics

• Accuracy: Depends on training data quality
• Processing Speed: Real-time capability
• Memory Usage: Optimized for standard hardware

📊 Sample Output

Compliance Detection

✅ HELMET DETECTED - COMPLIANT
Status: SAFE ✓
Confidence Level: 89.2%

Violation Alert

❌ NO HELMET DETECTED - VIOLATION
Status: UNSAFE ⚠️
Confidence Level: 94.7%

🚨 SAFETY VIOLATION ALERT
Immediate Actions Required:
- 🛑 Stop work immediately  
- 🪖 Provide safety helmet
- 📋 Brief worker on safety protocols
- 📝 Document the incident

🏗️ Project Structure

helmet_streamlit_app/
├── model/                   # Model directory
│   ├── model.json          # Model architecture in JSON format
│   ├── metadata.json       # Model metadata and configuration
│   └── weights.bin         # Model weights in binary format
├── app.py                  # Main Streamlit application
├── requirements.txt        # Python dependencies
├── labels.txt             # Class labels for model
├── README.md              # Project documentation
├── .gitignore            # Git exclusion rules
└── model.h5              # Complete Keras/TensorFlow model file

🔒 Security & Privacy

• Local Processing: All detection happens on your local machine
• No Data Upload: Images are not sent to external servers
• Privacy Compliant: Suitable for sensitive workplace environments
• Secure Logging: Violation logs stored locally

🚀 Deployment Options

Local Deployment

• Run on local machine for single-user access
• Ideal for testing and small-scale monitoring

Network Deployment

• Deploy on internal server for multi-user access
• Access from multiple devices on same network

Cloud Deployment (Advanced)

• Deploy on Streamlit Cloud, Heroku, or AWS
• Requires model hosting solution (Google Drive, etc.)

🛠️ Troubleshooting

Common Issues

Model Loading Error

❌ model.h5 file not found!

Solution: Ensure model.h5 is in the project root directory

Camera Access Error

Permission denied for camera access

Solution: Grant camera permissions in browser settings

Low Detection Accuracy

Many false positives/negatives

Solution: Adjust detection threshold or retrain model with better data

Performance Optimization

• Use GPU-enabled TensorFlow for faster processing
• Optimize image resolution for speed vs accuracy balance
• Consider model quantization for mobile deployment

🤝 Contributing

We welcome contributions to improve the system!

How to Contribute

1. Fork the repository
2. Create a feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

Areas for Contribution

• Model accuracy improvements
• Additional safety equipment detection
• Mobile app development
• Integration with IoT devices
• Advanced analytics features

📄 License

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

📞 Support & Contact

Getting Help

• Issues: Report bugs via GitHub Issues
• Discussions: Join project discussions
• Email: [harshitbhalani187@gmail.com]

FAQ

Q: Can this detect other safety equipment? A: Currently focused on helmets, but can be extended for vests, gloves, etc.

Q: What accuracy can I expect? A: Depends on training data quality, typically 85-95% with good data.

Q: Can it work with IP cameras? A: Yes, with minor code modifications for RTSP streams.

Q: Is it suitable for outdoor use? A: Yes, but performance may vary with lighting conditions.

🙏 Acknowledgments

• TensorFlow Team for the amazing ML framework
• Streamlit for the intuitive web app framework
• OpenCV for computer vision capabilities
• Google Teachable Machine for accessible model training

📈 Future Enhancements

• Multi-person detection in single image
• Integration with existing security systems
• Mobile application development
• Advanced analytics and predictive insights
• Support for multiple safety equipment types
• Real-time video stream processing
• Database integration for enterprise use
• API development for third-party integrations

---

⚡ Automated Safety Monitoring for a Safer Workplace ⚡

Made with ❤️ for workplace safety