A C++ Modular, Real-Time, and User-owned Surveillance System.

---

What is OpenGuard?

OpenGuard is a real-time home surveillance system written in C++ with a focus on modularity, efficiency, and privacy. It includes:

• Motion detection (MOG2)
• AI object detection (YOLOv5)
• Native + Python hook system for automation
• Real-time recording with metadata & compression
• Web dashboard and Telegram integration

includes a simple Python API to let users script automations without having to dive into the C++ internals. Windows support is tested and stable. Linux support is in progress, the code is cross-platform by design and just needs some testing.

---

Screenshots & Demo

Showcase

• Motion detection with region of interest
• Telegram bot control & alerts
• Live view via WebSockets
• Custom script hooks in Python
• Schedule system and pause/resume logic
• Event-based recording
• Tested with YOLOv5n

Program structure

---

Why?

Today's surveillance systems are often cloud-dependent and invasive by offloading footage to third-party servers. OpenGuard was built to challenge that model.

• Everything runs locally.
• From the C++ core to the Python hook system, every part is designed to be extended.
• Real-time and efficient.

OpenGuard is designed for users that want total control over their surveillance system.

---

Limitations & Future Work

• Currently only tested on Windows
• The web dashboard works but could use some UI cleanup
• The web code is not up to my code quality standards as it had to be rushed due to time constraint.
• A few more hook examples and tutorials (i.e. Trigger lights, check network for mac address to disable the system when user is home, etc..).

Todo

• /!!!\ Encrypt login ws -> OpenGuard.
• Better UI/UX for the panel
• Object-specific fine-tuned YOLO models

---

Building / Running

Was built with CMAKE and MSVC. Replace OpenCV path in CMAKE to compile. Release is provided with pre-compiled executable.

to run website. web/front/config: npm install

web/front/config: npm run dev

web/back: npm install

web/back: node server

py setup.py

Credits

Thank you to the maintainers of the following:

• nlohmann/json
• Asio
• WebSocket++
• base64.hpp by Tobias Locker