PPE Detection with YOLO

This project detects Personal Protective Equipment (PPE) in videos using a YOLO model and OpenCV.

Dataset

• Construction Site Safety Image Dataset (Roboflow) on Kaggle

Download the dataset from the above link and prepare your training data as needed.

Requirements

• Python 3.8+
• ultralytics
• opencv-python
• cvzone
• numpy

Install dependencies with:

pip install ultralytics opencv-python cvzone numpy

Usage

1. Download and extract the dataset from the Kaggle link above.
2. Place your video file in the video/ directory (e.g., video/ppe-3-1.mp4).
3. Make sure your trained YOLO model weights are in the correct path (e.g., runs/detect/train2/weights/best.pt).
4. Edit PPE.py if you want to change the video or model path.
5. Run the detection script:

python PPE.py

• The script will display the video with bounding boxes and labels for detected PPE items.
• Press q to quit the video window.

Notes

• If you see errors about missing modules, please make sure you have activated the correct Python environment and installed all requirements.
• If you see errors about missing files, check that the video and model paths in PPE.py are correct.

Project Structure

PPE detection/
├── PPE.py
├── video/
│   └── ppe-3-1.mp4
├── runs/
│   └── detect/
│       └── train2/
│           └── weights/
│               └── best.pt
├── README.md