Absolutely! Here’s the clean, corrected, and pastable version of your GitHub README:
FlySpy – AI-Enabled FPV Spy Drone
Open-source FPV AI Spy Drone Platform for Reconnaissance and Research A modular, cost-effective drone build with real-time video streaming, person detection, and BLE telemetry—designed for teaching, experimentation, and field deployment.
🔍 About the Project
This project was developed by the FlySpy group as part of the Master’s Project – Intelligent Systems (SS2025) at Frankfurt University of Applied Sciences.
🎯 Objective
Investigate and document a low-cost, AI-capable FPV drone system suitable for education, research, and real-world reconnaissance scenarios.
🧩 Project Scope
- Assemble a 5-inch FPV drone using accessible components
- Integrate lightweight, real-time person detection using MobileNet-SSD
- Display GPS and orientation telemetry from the drone using BLE
- Design a browser-based dashboard for video and sensor fusion
- Ensure reproducibility, modularity, and future scalability
🔧 Reference Build Components
| Component | Description |
|---|---|
| Frame | FlyFishRC Volador II VX5 O3 FPV Freestyle T700 Frame Kit (5 inch) |
| Flight Controller (FC) | SpeedyBee F405 AIO 40A Bluejay (3–6S) |
| Video Transmitter (VTX) | SpeedyBee TX800 Analog FPV VTX |
| Camera | Caddx Ratel Pro Analog FPV Camera (1500TVL) |
| Receiver | SpeedyBee Nano 2 2.4GHz ELRS |
| Remote Controller | Radiomaster BOXER + battery |
| Charger | SkyRC S100 Neo (1–6S LiPo, 10A, 100W AC) |
| Smoke Stopper | TBS Smoke Stopper 2–8S |
| FPV Goggles (for stream) | Skyzone Cobra X V4 with RCA video out (for AI integration via capture card) |
🚁 Getting Started
🛠️ Hardware Setup
- Assemble the frame and install motors, ESCs, and the flight controller
- Connect the ELRS receiver and verify power safety with a smoke stopper
- Power the system with a 3–6S LiPo battery
- Pair the drone with FPV goggles (e.g., Skyzone Cobra X V4 with RCA-out)
⚙️ Firmware Configuration
- Flash Betaflight (initially) or ArduPilot (for advanced GPS/autonomous features)
- Configure receiver protocol, motor direction, video output, and PID tuning
- Set failsafe behavior and assign flight modes
📺 Viewing the FPV Stream on Mac/Linux
- RCA video output from FPV goggles → RCA-to-USB capture card → MacBook
- View the stream using OBS or QuickTime via UVC input
🧠 AI Integration
🎯 Goal
Enable real-time person detection on the FPV stream to simulate reconnaissance and surveillance capabilities.
🔧 Detection Pipeline
- Input: Analog video via USB capture card
- Model: MobileNet-SSD (via OpenCV
cv2.dnn) - Processing: Person detection and ID tracking on each frame
- Output: Live video with bounding boxes and ID overlays
🖥️ Dashboard Features
We built a browser-based dashboard using Flask, OpenCV, HTML/CSS, and SocketIO to:
- Stream video and overlay AI detection
- Display live BLE-based telemetry (GPS, orientation)
- Scan for and connect to nearby drones
- Show heading via compass and location via map
- Operate fully in-browser (tested on Chrome and Safari)
# Setup example
python3 -m venv venv
source venv/bin/activate
pip install flask flask-socketio opencv-python bleak eventlet
python app.py
Let me know if you’d like to add sections like System Architecture, Screenshots, License, or Contribution Guidelines!