SOP.md

🛡️ Smart Live Surveillance Using YOLOv8

Overview

This project implements a real-time smart surveillance system using the latest YOLOv8 object detection model and Gradio for live webcam streaming. It can detect any type of object, label it in real-time, record video streams, and generate Excel reports with detected objects and timestamps.

It’s fully compatible with Hugging Face Spaces for cloud deployment.

---

Features

• Live Object Detection – Detects and labels objects in real-time using YOLOv8.
• Excel Reporting – Saves detected objects with timestamp, name, and confidence score.
• Video Recording – Option to record annotated video with a simple toggle.
• Hugging Face Deployment – Fully compatible with Spaces and no manual credentials needed.
• Cross-Platform – Works on Windows, Linux, or cloud environments.

---

Technology Stack

• Ultralytics YOLOv8 – Real-time object detection framework.
• Gradio – Web UI for live webcam streaming and controls.
• OpenCV – Video and image processing.
• Pandas + OpenPyXL – Excel report generation.
• Hugging Face Hub – Model hosting and deployment.

---

Installation

1. Clone this repository:

git clone https://huggingface.co/Michtiii/live_surveillance
cd live_surveillance

2. Install dependencies:

pip install -r requirements.txt

---

Usage

1. Launch the Gradio interface:

python live_surveillance.py

2. Features in the interface:

• Live Camera Feed – Displays annotated video in real-time.
• Record Video Checkbox – Start/stop recording the live feed.
• Download Report Button – Download Excel report of detected objects.

---

Directory Structure

live_surveillance/
│
├── live_surveillance.py     # Main application code
├── requirements.txt        # Project dependencies
└── detection_report.xlsx   # Generated Excel report (runtime)

---

Deployment on Hugging Face Spaces

1. Create a new Space (Gradio type).
2. Upload live_surveillance.py and requirements.txt.
3. Click “Deploy” and wait for build to complete.
4. Open the Space to use live object detection.

---

Notes

• Uses sources=["webcam"] for live streaming (Gradio v6+ compatible).
• YOLO configuration files are automatically stored in /tmp/Ultralytics.
• Hugging Face Hub handles YOLOv8 model download automatically.
• Video recordings are saved temporarily; you can modify the path for permanent storage.

---

Future Enhancements

• Multiple camera streams
• Alerts via email/SMS for specific objects
• Real-time dashboard for object counts and trends
• Integration with cloud storage for reports and video

---

License

MIT License – Feel free to use, modify, and deploy this project. kajaldadas149@gmail.com