VisionStack_PPEDetection

1. Project Title

• Application Name: PPE Kit Detection
• Short Description: A Flask-based application that utilizes OpenCV for real-time detection of Personal Protective Equipment (PPE) on individuals.

2. Overview

• Purpose: - To detect if a person is wearing the Personal Protective Equipments (PPE kits) while working on the Factory Floor.
• Functionality: - PPE kit detection using the YOLO v8 algorithm from the Ultralytics library. - The model is able to track if a person (Factory Worker) is wearing safety equipments like Helmet,Safety Vest, Safety Boots, Glasses and Gloves. - The model highlights the equipment wore by the person in a green colour and displays a text indicating the equipment. - The model displays the equipment only after a certian level of confidence which is dynamic and can be changed according to the need. - The model is deployed on localhost on port 5000 using the Flask API.

3. Installation Instructions

• Prerequisites: - Python version 3 - Flask version 3 - OpenCV-Python library

• Installation Steps: - Clone/download the repository - Set up a Python virtual environment - Install required Python packages: pip install -r requirements.txt

4. Usage

• Instructions to start the server: - Run the flask_PPE.py for Video stream or flask_PPE_Photo.py for a single frame from the Video stream (or photo). - Copy the IP address from the terminal and paste it into a browser window and add /video_feed/<item> here replace /<item> with the safety equipment you want to check for (ex Helmet) and for checking for all the equipments at once replace /<item> with /all - For a single frame instead of /video_feed/<item> do /photo/<item>
• Output: - The output of the model can be viewed on a browser in a video format and on the terminal in text format. - The output window "Frame" consists of the original video plus the detected items from a safety[] list, their confidence level, Number of people and the Frame Rate of the application. - Same items can also be viewed in the terminal.

5. Code Structure

• Modules and Packages: Description of the project's directory structure and an overview of the main modules and packages. - flask_PPE.py and flask_PPE_Photo.py - The main Flask application files. - PPE_rec.py and PPE_rec_p.py - Modules for PPE detection using OpenCV. - templates/ - Folder containing HTML files for the web interface.

6. API Reference

• Endpoints: - /video_feed/<item> and /photo/<item> - Endpoint for processing images and returning detection results.

6. Performance
   • The average FPS for AMD Ryzen 5, GTX 1650, 16 GB RAM is 50 FPS
   • The average FPS for Jetson Nano is FPS