This repository presents a solution for Real-Time Personal Protective Equipment (PPE) Detection and Tracking in manufacturing environments. The project leverages CCTV cameras 📹 to monitor workers and ensure safety compliance through the detection of essential PPE like helmets 👷, vests 🦺, and head🧑🏭. By utilizing cutting-edge technologies like YOLOv8 🔥 and BotSORT 🧠, along with custom-developed algorithms 💡, this system is capable of accurately detecting and tracking multiple workers' PPE statuses in dynamic environments with high speed ⚡ and accuracy 🎯.
This project contributes to the Information Technology and Quality Engineering (ITQE) field by offering a scalable, real-time solution for industrial safety monitoring. With an extensive dataset of 54,325 annotated images 📸 of helmets, vests, and heads, the model is designed by authors to achieve high accuracy while ensuring that it works efficiently in real-time environments.
- YOLOv8: A state-of-the-art object detection model that has been customized for PPE detection tasks 🤖
- BotSORT: A robust tracking algorithm used to track multiple objects in real-time, even in crowded environments in workplace🏭🚶♂️🚶♀️
- Custom Algorithms: Proprietary algorithms have been developed to enhance the accuracy and performance of the system, ensuring that it runs fast and efficiently in real-time 🧠💨
- Python: The core programming language used for the implementation and training model🐍
- OpenCV: Used for image processing and video streaming from CCTV cameras 🖥️
- PyTorch: Deep learning framework for model training and inference 🔥
The project utilizes a substantial dataset with a total of 54,325 images (JPG/PNG) containing various instances of:
- Helmets 👷
- Vests 🦺
- Heads 🧑🏭
This dataset has been annotated to train and test the model’s ability to detect and classify PPE in a variety of industrial settings, ensuring that the system can generalize well across different scenarios.
Asking access to the DATASET : [email protected] 📧
input_video_path: Path to the input video (CCTV footage) 🎥
output_video_path: Path to save the output video with tracked objects 🎬
This repository also serves as the basis for an academic research paper 📄 in the field of Information Technology and Quality Engineering (ITQE), addressing critical issues related to worker safety in manufacturing. The paper explores:
-
The application of real-time computer vision for safety monitoring 🔍
-
The challenges and solutions related to tracking multiple objects (workers) across different environments ⚙️
-
The potential of AI-driven systems 🤖 to improve workplace safety and compliance ⚖️
-
Real-Time Detection: Instantly identifies and tracks PPE usage in live CCTV streams ⏱️
-
Multiple Object Tracking: Handles tracking of multiple workers simultaneously using BotSORT 👷♂️👷♀️
-
Custom Model & Algorithms: A tailored version of YOLOv8, optimized specifically for PPE detection and enhanced with proprietary algorithms for fast and high-accuracy real-time performance 🧠
-
Scalability: Suitable for large-scale deployment in industrial environments with multiple CCTV cameras 🏗️
- Provides continuous, real-time monitoring of worker safety
⚠️ - Seamlessly tracks multiple workers without losing accuracy or performance 🏃♂️🏃♀️
- High-Speed Processing: Optimized algorithms allow the system to run in real-time without lag or delay ⚡
This project is licensed under the MIT License - see the LICENSE file for details 📃
We welcome contributions from the community 🌍. To contribute, please follow these steps:
For any inquiries related to this project or collaboration, feel free to reach out:
Email: [email protected] 📧
GitHub: azimjaan21 👨🏻💻