AI Security Cam - Jetson Nano

📌 Overview

Sentinel Edge is a high-performance, containerized object detection solution designed for real-time video analytics directly on edge devices (e.g., Nvidia Jetson Nano, Google Coral TPU, or similar). It runs YOLOv5 for object detection, provides a live web dashboard, supports on-device alerting, and includes a robust OTA update system with CI/CD integration for seamless deployments.

This project minimizes bandwidth usage and latency by processing video locally, sending only relevant alerts or clips to the cloud.

🚀 Key Features

🎯 Real-Time Object Detection

• Runs YOLO model optimized with TensorRT on the Jetson Nano.
• Detects custom-trained objects (e.g., people, weapons, intrusion events).
• Works with USB cameras, IP cameras, or CSI cameras.

🌐 Web Dashboard

• Live video stream with detection overlays.
• Mobile-friendly UI built with Bootstrap 5.

🔄 OTA Updater

• Pulls new Docker images automatically from a container registry.
• Performs health checks before switching to the new version.
• Automatic rollback if update fails.

⚙️ CI/CD Pipeline

• GitHub Actions builds Docker images on each commit.
• Pushes images to a cloud container registry (Google Artifact Registry or Docker Hub).
• Deploys new builds to Jetson Nano automatically.

⚡ Hardware Acceleration

• NVIDIA Jetson Nano GPU acceleration.
• TensorRT engine conversion for faster inference.
• Optimized to run 24/7 with low power consumption.

🔒 Offline Operation & Alerting

• Can work fully offline for local-only installations.
• Saves last 30 seconds of video for every detection event.
• Supports push notifications or cloud API calls for alerts.

🔥 Thermal & Power Management

• Active thermal monitoring.
• Fan control or throttling for long-term reliability.

---

🛠️ System Architecture

The architecture is composed of three primary containers:

1. OTA Updater — Handles updates, rollback, and version control.

2. App Container — Runs camera feed processing, YOLO detection, and alert handling.

3. Web Container — Serves live video and detection data via a browser dashboard.

📂 Project Structure

├── sentinel_app/       # YOLO detection and alert logic
├── sentinel_web/       # Web dashboard source code
├── sentinel_src/       # Source Files that should put in Jetson Nano 
|   └── ota_updater/    # OTA update service
├── images
|   └── schematic.png   # System architecture diagram
└── README.md           # Main project documentation

⚙️ Tech Stack

• Hardware: Nvidia Jetson Nano / Google Coral TPU
• ML Model: YOLOv5
• Languages: Python, JavaScript
• Frameworks: Flask (Web API), Bootstrap (Frontend)
• Containerization: Docker
• CI/CD: GitHub Actions
• Streaming: WebSocket

📦 Installation & Setup

1. Clone the Repository

git clone https://github.com/RajWasarwad/ai-security-cam

2. Run the Main file

cd sentinel_src
sudo ./main.bash

📜 License

MIT License © 2025 InGen Dynamics Inc.

📧 Contact

For inquiries, support, or contributions:

wasarwad.raj@gmail.com