SafeZone AI – Real-Time Safety Detection System

Project Overview

SafeZone AI is an innovative solution designed to enhance safety in industrial environments, particularly in companies dealing with chemical products. This system leverages artificial intelligence for real-time detection of personal protective equipment (PPE) and identification of authorized personnel in sensitive areas. The main objective is to prevent incidents by immediately alerting in case of non-compliance or unauthorized intrusion.

Client Context

This project addresses the needs of a Tunisian industrial company aiming to strengthen on-site safety. The key requirements include:

• Automated Monitoring: Use of existing surveillance cameras for continuous detection.
• Anomaly Detection: Automatic identification of missing PPE (helmet, vest, glasses) and unauthorized individuals.
• Centralized Dashboard: A dashboard to visualize real-time alerts and track statistics by day and by area.
• Local AI Processing: AI processing is done locally or on a private server to ensure high data confidentiality, avoiding external cloud services.

Key Features

• PPE Detection: Automatic recognition of helmet, safety vest, and safety glasses.
• Facial Recognition: Identification of authorized individuals in restricted areas.
• Alert System: Instant alert generation in case of safety rule violations or unauthorized access.
• Interactive Dashboard: Visualization of safety data, including number of alerts per day, violation videos, and list of unrecognized individuals.

Technical Architecture

SafeZone AI is built with several interconnected components, ensuring robust performance and high confidentiality.

AI Components

• PPE Detection: Uses YOLOv8 for fast and accurate PPE detection.
• Facial Recognition: Combines MTCNN (Multi-task Cascaded Convolutional Networks) for face detection and FaceNet for generating facial embeddings and recognition.
• Face Database: A SQLite database stores embeddings of authorized employee faces, enabling fast and secure verification.

Backend

The backend is developed using Flask, a lightweight Python web framework, to handle server-side logic and interactions with AI models and the database.

• Video Stream: Uses Socket or WebRTC for real-time video transmission from surveillance cameras.
• REST API: A RESTful API is provided for accessing alert data and statistics, enabling integration with other systems or applications.

Frontend

The interactive dashboard is built using Streamlit or Gradio (in this implementation, an HTML/CSS/JS frontend is used).

• Alert Visualization: Displays number of alerts per day for quick incident monitoring.
• Violation Videos: Access to video recordings showing moments of safety rule violations.
• Unrecognized People List: Identification of unauthorized individuals entering sensitive areas.

Storage

• Database: SQLite (or PostgreSQL for larger deployments) is used to store encoded faces, alert logs, and history of videos (or links to local files).

Future Enhancements

Possible future improvements include integration with real-time alert systems via Twilio (for SMS/calls) or a Telegram Bot to instantly notify the security team of incidents.

Dashboard Screenshots

Below are some previews of the SafeZone AI dashboard, showcasing its monitoring and reporting features.

Overview

Equipment Statistics

Installation and Usage

Prerequisites

• Python 3.x
• pip (Python package manager)
• OpenCV (for video processing)

Installation Steps

1. Clone the repository:

   git clone https://github.com/your_user/SafeZone-AI.git
   cd SafeZone-AI

2. Install Python dependencies:

   pip install -r requirements.txt

3. Prepare AI Models:

   Ensure the YOLOv8 model (best.pt) is placed in the models/ folder.

4. Prepare the Face Database:

   Place images of authorized personnel in the data/known_faces/ directory.

Launching the Application

1. Start the Flask backend:

   python src/app.py

2. Access the dashboard:

   Open your browser and go to http://localhost:5000/dashboard for the dashboard or http://localhost:5000/ for the live video stream (if the camera is set up).

Project Structure

SafeZone-AI/
├── src/
│ ├── app.py # Main Flask application
│ ├── detect.py # YOLOv8 detection logic
│ ├── detect_face.py # Face detection and recognition
│ ├── detect_only_faces.py # Face detection only
│ ├── face_db.py # Face database management
│ ├── log.py # Alert logging
├── frontend/
│ ├── dashboard.html # HTML template for dashboard
│ ├── index.html # HTML template for homepage (video stream)
│ ├── script.js # JavaScript for frontend interactions
│ ├── style.css # CSS styles for the frontend
├── models/
│ ├── best.pt # Trained YOLOv8 model
├── data/
│ ├── known_faces/ # Images of authorized individuals
│ ├── log.csv # Alert logs file
├── notebooks/
│ ├── model.ipynb # Jupyter notebook for training/experiments
├── docs/
│ ├── images/ # Documentation screenshots and images
├── requirements.txt # Python dependencies
├── README.md # This file
├── .gitignore # Git ignore rules

Contribution

Contributions are welcome! Please follow these steps:

1. Fork the repository.
2. Create a new branch (git checkout -b feature/new-feature).
3. Make your changes and commit them (git commit -am 'Add new feature').
4. Push to the branch (git push origin feature/new-feature).
5. Create a Pull Request.

License

This project is licensed under the MIT License. See the LICENSE file for more details.

Contact

For any questions or suggestions, please open an issue on this GitHub repository.