Workplace Safety PPE Detection using YOLOv8

DEMO: https://huggingface.co/spaces/hafizqaim/Workspace-Safety-Detection

This project provides a real-time object detection system to monitor whether people in a workplace environment are wearing essential Personal Protective Equipment (PPE), specifically safety helmets and vests. The system is built using the YOLOv8 model.

---

Key Features

• Real-Time Detection: Capable of processing video streams from files or a live webcam to identify PPE in real-time.
• High Accuracy: Trained on a large dataset of over 23,000 images, achieving an mAP50 of 73.5% overall and over 86% for key classes like helmets and vests.
• State-of-the-Art Model: Utilizes YOLOv8, a powerful and efficient object detection architecture.
• Focused Detection: While trained on 17 classes, the inference script is configured to specifically highlight helmets and vests for workplace safety monitoring.

---

Technology Stack

• Python 3.8+
• PyTorch
• Ultralytics YOLOv8
• OpenCV
• Kaggle Notebooks (for training)

---

Getting Started

Follow these instructions to set up and run the project on your local machine.

Prerequisites

• Python 3.8 or newer
• Git

Installation

1. Clone the repository:

   git clone [https://github.com/hafizqaim/Workspace-Safety-Detection-using-YOLOv8.git](https://github.com/hafizqaim/Workspace-Safety-Detection-using-YOLOv8.git)
   cd Workspace-Safety-Detection-using-YOLOv8

2. Create and activate a virtual environment:

   # For macOS/Linux
   python3 -m venv venv
   source venv/bin/activate
   
   # For Windows
   python -m venv venv
   .\venv\Scripts\activate

3. Install the required packages:

   pip install -r requirements.txt

---

Usage

1. Download the Trained Model: The trained model file (best.pt) is required to run the inference. Download it from the Releases page of this repository.

2. Place the Model: Place the downloaded best.pt file in the root directory of the project.

3. Run the Inference Script: The inference.py script is configured to run on your webcam by default.

   python inference.py

   • To use a video file instead, open inference.py and modify the script to point to your video file.

---

Model Performance

The model was trained for 10 epochs on the "PPE Detection v3" dataset from Roboflow.

Class	Precision	Recall	mAP50	mAP50-95
Overall	0.72	0.715	0.735	0.456
head_helmet	0.784	0.824	0.866	0.584
vest	0.841	0.897	0.935	0.705