🔥 TrueSight — AI-Based Safety Equipment Detection

YOLOv8 + FastAPI + React | Full Stack Computer Vision System

---

🚀 Overview

TrueSight is an AI-powered computer vision system designed to detect safety equipment in images.
It uses a custom-trained YOLOv8s model, a FastAPI backend, and a beautiful React web interface.

This project was built for the HackBeyond Hackathon 2025.

---

🎯 Features

• ✔ Detect 7 safety equipment classes
• ✔ Real-time image upload & prediction
• ✔ Annotated image returned (Base64)
• ✔ MERN-style frontend (React + API backend)
• ✔ Fully Dockerized backend for smooth deployment
• ✔ mAP50 score: 0.78
• ✔ Simple & fast REST API for integration

---

🧠 Model Details

Model Used: YOLOv8s

7 Classes:

• Oxygen Tank
• Nitrogen Tank
• First Aid Box
• Fire Alarm
• Safety Switch Panel
• Emergency Phone
• Fire Extinguisher

Training Stats

Metric	Value
Epochs	50
mAP50	0.787
mAP50-95	0.661
Precision	0.90
Recall	0.72

Confidence Threshold

We use:

0.25 (25%)

To eliminate low-confidence predictions and reduce false positives.

---

📁 Project Structure

TrueSight/
│
├── backend/
│   ├── server.py
│   ├── Dockerfile
│   ├── requirements.txt
│   ├── models/
│   │    └── best.pt
│   └── utils/
│
└── frontend/
    ├── src/
    │   ├── App.jsx
    │   └── styles.css
    ├── public/
    └── package.json

---

🧪 API Documentation

POST /predict

Upload an image → get back annotated image + detections.

Request

POST /predict
Content-Type: multipart/form-data
file: image.jpg
conf: 0.25

Use Our Webapp

---

🐳 Backend Setup (Local)

Install dependencies

cd backend
pip install -r requirements.txt

Run server

.\.venv\Scripts\Activate
python -m uvicorn server:app --host 0.0.0.0 --port 8000 --workers 1

---

🌐 Frontend Setup (React)

Install dependencies:

cd frontend
npm install

Run Dev Server:

npm start

---

🧾 Hackathon Submission Summary

Submission Item	Status
Model trained	✔ Done
mAP50 score	0.78
best.pt	✔ Provided
Working web app	✔ Deployed
GitHub repository	✔ Public
Slides / PPT	✔ Attached
Demo video (optional)	Pending

🙏 Acknowledgements

• Ultralytics YOLOv8
• FastAPI Community
• ReactJS Community
• HackBeyond Hackathon

---

⭐ If you like this project

Give the repo a star ⭐ on GitHub!