---

YOLOv8-Based Person and PPE Detection

This project involves training YOLOv8 models to detect persons and Personal Protective Equipment (PPE) in images. The project includes format conversion, model training, and inference.

Project Overview

• Person Detection Model: A YOLOv8 model trained to detect persons in images.
• PPE Detection Model: A YOLOv8 model trained on cropped images of detected persons to identify various PPE items (e.g., hard-hats, gloves, boots).

Steps to Run the Project

1. Format Conversion

Convert PascalVOC annotations to YOLOv8 format:

python pascalVOC_to_yolo.py /path/to/pascalVOC_labels /path/to/output_dir

Note : The directory containing .xml files should have exact name 'pascalVOC_labels'

2. Model Training

Train the person detection and PPE detection models using YOLOv8. Use the provided training scripts and datasets.

3. Inference

Run inference on a directory of images:

python inference.py --input_dir /path/to/input/images --output_dir /path/to/output/images --person_det_model /path/to/person/model --ppe_detection_model /path/to/ppe/model

Requirements

Install the necessary packages using:

pip install -r requirements.txt

Project Structure

• pascalVOC_to_yolo.py: Script for converting PascalVOC annotations to YOLOv8 format.
• inference.py: Script for running inference with both person and PPE detection models.
• requirements.txt: List of required Python packages.